I will briefly comment Carson et al interpretations of the origin of the effect they demonstrate. I wish to attract the attention to the fact that synchronization, and in general coordination, is dependent, even under rather drastic simplifications, upon various properties of components and more widely of coordination, some being considered by the authors, and more generally in the recent literature, others not. How can the stability of synchronization be improved? One may think first by reducing inherent random variability of the movement caused by noise present in the nervous system, especially at coincidence points with the external events. This may be achieved in two ways. Firstly as considered by the authors by minimizing the amplitude of noise, considering for instance the signal dependent noise hypothesis (Schmidt et al 1979; Harris and Wolpert, 1998). This may amount to reducing the force produced to move. In short, in the present case this means going with gravity.

Secondly one may use co-activation to counteract the noise; conceptually this corresponds to increasing the stability of the joint motion. The interplay between signal dependent noise and this type of stability, interpreted by some as impedance control, gives rise to the actual movement variability, in some case the stability gained by co-contraction can overcome the accompanying increase of noise (see Selen et al 2005; Selen et al 2009). In stochastic dynamical systems this sort of problem may be generally formulated as a Langevin equation. Stability is provided by a drift term and noise by a diffusion term, in a signal dependent case the noise being considered as multiplicative. When applied to oscillating movement, damping terms and stiffness (see Kay et al, 1987) can act to stabilize the motion against noise. This second option doesn’t seem to be used here, to the contrary, as passive movement is preferred.

Another interesting aspect which is not considered here is not directed at reducing the joint variability per se but at improving the impact of coupling with external events on the movement. If the movement is very stable, say by impedance control or increase of damping, a coupling for synchronization with another effector or with perceived external events will hardly influence its movement. Indeed coupling is also, like noise, considered as a perturbation of the component’s dynamics. Sensitivity to coupling, in the present case mediated by neural networks integrating auditory, timing and movement planning cortical and sub- cortical areas, is inversely related to the stability of the joint. Hence going with gravity, by reducing muscular contraction, may increase the sensitivity to this coupling and promote a synchronization which is preferentially selected by the CNS. Of course such processes remain hypothetical, and minimally the functioning of coupling and joint stability should be framed in a coherent, maybe multilevel, description.