Hadjiosif AM & Smith MA (2015) Flexible Control of Safety Margins for Action Based on Environmental Variability.
Journal of Neuroscience, 35(24) 9106-9121 (June 17, 2015).
Abstract: To reduce the risk of slip, grip force (GF) control includes a safety margin above the force level ordinarily sufficient for the expected load
force (LF) dynamics. The current view is that this safety margin is based on the expected LF dynamics, amounting to a static safety factor
like that often used in engineering design. More efficient control could be achieved, however, if the motor system reduces the safety
margin when LF variability is low and increases it when this variability is high. Here we show that this is indeed the case by demonstrating
that the human motor system sizes the GF safety margin in proportion to an internal estimate of LF variability to maintain a fixed
statistical confidence against slip. In contrast to current models of GF control that neglect the variability of LF dynamics, we demonstrate
that GF is threefold more sensitive to the SD than the expected value of LF dynamics, in line with the maintenance of a 3-sigma confidence
level.Wethen show that a computational model of GF control that includes a variability-driven safety margin predicts highly asymmetric
GF adaptation between increases versus decreases in load. We find clear experimental evidence for this asymmetry and show that it
explains previously reported differences in how rapidly GFs and manipulatory forces adapt. This model further predicts bizarre nonmonotonic
shapes for GF learning curves, which are faithfully borne out in our experimental data. Our findings establish a new role for
environmental variability in the control of action.