Effects of neuromuscular lags on controlling contact transitions.

Abstract

We present a numerical exploration of contact transitions with the fingertip. When picking up objects our fingertips must make contact at specific locations, and—upon contact—maintain posture while producing well-directed force vectors. However, the joint torques for moving the fingertip towards a surface (τm) are different from those for producing static force vectors (τf). We previously described the neural control of such abrupt transitions in humans, and found that unavoidable errors arise because sensorimotor time delays and lags prevent an instantaneous switch between different torques. Here, we use numerical optimization on a finger model to reveal physical bounds for controlling such rapid contact transitions. Resembling human data, it is necessary to anticipatorily switch joint torques to τf at about 30 ms before contact to minimize the initial misdirection of the fingertip force vector. This anticipatory strategy arises in our deterministic model from neuromuscular lags, and not from optimizing for robustness to noise/uncertainties. Importantly, the optimal solution also leads to a trade-off between the speed of force magnitude increase versus the accuracy of initial force direction. This is an alternative to prevailing theories that propose multiplicative noise in muscles as the driver of speed–accuracy trade-offs. We instead find that the speed–accuracy trade-off arises solely from neuromuscular lags. Finally, because our model intentionally uses idealized assumptions, its agreement with human data suggests that the biological system is controlled in a way that approaches the physical boundaries of performance.