Does dynamic stability govern propulsive force generation in human walking?

Abstract

Before succumbing to slower speeds, older adults may walk with a diminished push-off to prioritize stability over mobility. However, direct evidence for trade-offs between push-off intensity and balance control in human walking, independent of changes in speed, has remained elusive. As a critical first step, we conducted two experiments to investigate: (i) the independent effects of walking speed and propulsive force (F_P) generation on dynamic stability in young adults, and (ii) the extent to which young adults prioritize dynamic stability in selecting their preferred combination of walking speed and F_P generation. Subjects walked on a force-measuring treadmill across a range of speeds as well as at constant speeds while modulating their F_P according to a visual biofeedback paradigm based on real-time force measurements. In contrast to improvements when walking slower, walking with a diminished push-off worsened dynamic stability by up to 32%. Rather, we find that young adults adopt an F_P at their preferred walking speed that maximizes dynamic stability. One implication of these findings is that the onset of a diminished push-off in old age may independently contribute to poorer balance control and precipitate slower walking speeds.

Keywords: ageing; balance; biofeedback; push-off; variability; walking speed.

Figure 1.

Group mean (standard deviation) peak propulsive force values (a) when walking across a range of speeds and at 1.3 m s⁻¹ with biofeedback of propulsive force targets extracted from slower speeds and (b) walking at preferred speed with propulsive forces ±10% and ±20% larger than preferred. (c) Experimental design using visual biofeedback of real-time propulsive force values calculated from a dual-belt, force-measuring treadmill to decouple the effects of walking speed and propulsive force generation on metrics of dynamic balance control. Asterisks represent significant (p < 0.05) difference from prescribed values.

Figure 2.

Group mean (standard deviation) short-term maximum divergence exponents (λ_S) defined using anterior–posterior (AP), mediolateral (ML), vertical (Vert) and three-dimensional (3D) sacrum kinematics for (a) systematic changes in walking speed and propulsive force targets extracted from slower speeds, and (b) walking at preferred speed with propulsive forces smaller or larger than preferred. Double asterisks represent significant (p < 0.05) main effects of speed or propulsive force. ‘a’ Indicates significant (p < 0.05) pairwise difference at matched propulsive forces, ‘b’ indicates significantly (p < 0.05) different from normal walking with biofeedback, and ‘c’ indicates significantly (p < 0.05) different from normal walking without biofeedback.

Figure 3.

Group mean (standard deviation) anterior–posterior (AP), mediolateral (ML) and vertical (Vert) sacrum position variability for (a) systematic changes in walking speed and propulsive force targets extracted from slower speeds, and (b) walking at preferred speed with propulsive forces smaller or larger than preferred. Double asterisks represent significant (p < 0.05) main effects of speed or propulsive force. ‘a’ Significant (p < 0.05) pairwise difference at matched propulsive forces, ‘b’ significantly (p < 0.05) different from normal walking with biofeedback and ‘c’ significantly (p < 0.05) different from normal walking without biofeedback.

Figure 4.

Group mean (standard deviation) anterior–posterior (AP), mediolateral (ML) and vertical (Vert) sacrum velocity variability for (a) systematic changes in walking speed and propulsive force targets extracted from slower speeds, and (b) walking at preferred speed with propulsive forces smaller or larger than preferred. Double asterisks represent significant (p < 0.05) main effects of speed or propulsive force. ‘a’ Significant (p < 0.05) pairwise difference at matched propulsive forces, ‘b’ significantly (p < 0.05) different from normal walking with biofeedback and ‘c’ significantly (p < 0.05) different from normal walking without biofeedback.

Figure 5.

Group mean per cent change in peak propulsive force versus per cent change in step length from experiment 2, plotted against data adapted from Martin & Marsh [47]. We found that subjects increased (decreased) their step lengths by 6.6% (−8.0%) when targeting a 20% increase (decrease) in F_P, compared with walking normally. Conversely, Martin & Marsh [47] found that their subjects increased (decreased) F_P by only 12.2% (−13.0%) when directly increasing (decreasing) their step lengths by 10.2% (−7.9%). These results imply that, although both change simultaneously, modifying propulsive forces is not biomechanically equivalent to modifying step lengths. Accordingly, we interpret our findings to allude to direct effects of modulating push-off intensity and not solely a result of secondary changes in step length.