Shaking when stirred: mechanisms of physiological tremor

Point your finger at the full stop at the end of this sentence. If you look closely you will see that your finger does not remain perfectly still. Instead it exhibits small amplitude, high frequency oscillations. This is an example of physiological tremor.

Why do we wobble when we don't intend to?

Unsurprisingly, neurophysiologists have sought neurophysiological explanations. It has been proposed that physiological tremor is caused by oscillatory reflex loops which induce synchronous motor unit activity (Lippold, 1970; Christakos et al. 2006) or by the force fluctuations produced when the incompletely fused twitches of just-recruited motor units are low-pass filtered by the elastic properties of the muscle (Allum et al. 1978; Christakos et al. 2006). Other hypotheses involve oscillating networks in the central nervous system (Bye & Neilson, 2010).

A common feature of these explanations is that they invoke mechanisms which cause muscle force to fluctuate at frequencies similar to tremor frequencies (7–11 Hz). But that may not be necessary. It has long been known that fingers (Lippold, 1970) and hands (Lakie et al. 1986) have resonant frequencies that are very similar to their tremor frequencies. This simple observation has important implications for the understanding of tremor: it means that tremor need not be driven by a process that is dominated by tremor frequencies. Complex perturbations with broad frequency spectra could sustain tremor. It is not necessary, when seeking to explain tremor, to invoke mechanisms which cause muscle force to oscillate at 7–11 Hz.

In a recent issue of The Journal of Physiology, Lakie and colleagues (Lakie et al. 2012) describe investigations into the mechanisms of hand tremor. They began by recording wrist accelerations and EMG in wrist muscles while subjects held the wrist at a target angle. The frequency spectrum of the EMG was quite flat above 2 Hz but the frequency spectrum of the tremor (acceleration) had a distinct peak at 8 Hz. This is strongly suggestive of resonance.

In (nearly) static tasks, such as pointing or holding the wrist at a target angle, tremor is obvious. Less obvious, but equally intriguing, is that tremor also manifests during movement. When Lakie and colleagues measured hand tremor during a tracking task they found that the amplitude of the tremor was greater and the peak frequency less under these dynamic conditions compared to the static condition.

Why does the frequency and gain of tremor differ under static and dynamic conditions? Lakie and colleagues argue that the change in the frequency and gain is due to a change in the mechanical properties of muscles which traverse the wrist. When relaxed muscles are ‘stirred’ by subjecting them to large enough amplitudes of movement (greater amplitudes than occur in physiological tremor) they become much less stiff. This changes the resonance properties of the hand.

A computer model was used to test the idea that physiological tremor can be explained by resonance. The model consisted of a hand which rotated about a frictionless wrist joint. The wrist was spanned by a single muscle which behaved like a damped linear spring in series with a linear elastic tendon except that, in addition, the muscle generated active force. Active force was modelled by passing the measured, rectified EMG through a low-pass filter. Alternatively, active force was modelled by passing white noise through the same filter.

The simulations show that physiological tremor can be explained in entirely mechanical terms, without the need to invoke mechanisms that cause muscle force to fluctuate at tremor frequencies. When EMG was used as the model input, this simple model produced tremor-like acceleration profiles. Remarkably, it was not necessary to use real EMG to generate physiological tremor – white noise did the job just as well. This beautifully illustrates that random perturbations (rather than perturbations dominated by tremor frequencies) can sustain physiological tremor. The decrease in frequency and increase in amplitude that are observed in dynamic conditions were simulated simply by reducing the stiffness of the muscle. Realistic reductions in stiffness closely reproduced the decrease in frequency and increase in amplitude observed in experiments.

The model used here is clearly simplistic. Real wrists consist of many muscles, some acting antagonistically to others, each with their own complex passive mechanical properties, each generating unique patterns of motor unit activity, and each arranged in parallel and series with ligaments, fascia, skin and other tissues. But the strength of this model is in its simplicity. It shows that quite subtle features of physiological tremor can be explained without the need to invoke complex neural mechanisms. Lakie and colleagues’ hypothesis – that physiological tremor is primarily a mechanical, not a neurophysiological phenomenon – is satisfying because it is simple. It provides a unified, quantitative explanation of the characteristics of physiological tremor under static and dynamic conditions.

Where does this leave hypotheses about motor unit synchronisation and observations of coherence between motor unit firing and tremor? Interestingly, Lakie and colleagues speculate that these phenomena may be the consequences, rather than the causes, of tremors. Synchronised motor unit activation may be an epiphenomenon, rather than the primary mechanism of physiological tremor.