Balance-related outcomes.(a) Metabolic rate was reduced with Stabilizing control compared to Zero Gain and Destabilizing control conditions. For example, metabolic rate was 8.5% lower in the Stabilizing High Gain control condition than in the Destabilizing High Gain control condition (p = 0.02). Wearing the prosthesis increased metabolic rate, as did application of the disturbance. (b) Step width variability was lower with Stabilizing control than in Zero Gain or Destabilizing Gain conditions. Wearing the prosthesis appeared to increase step width variability, as did application of the disturbance. (c) Subjects appeared to prefer Stabilizing control conditions, although this trend was not statistically significant. Subjects preferred Normal Walking over wearing the prosthesis, and preferred not to have the random landing-angle disturbance. Blue bars correspond to Stabilizing control conditions, white bars to the Zero Gain condition, and red bars to Destabilizing conditions. Darker blue and red bars correspond to High Gains. Light gray bars correspond to the No Disturbance condition, and dark gray bars correspond to the Normal Walking condition. The p-values at top are for repeated measures ANOVA tests for an effect of control gain. Asterisks (*) indicate statistical significance among control gain conditions, and pluses (+) indicate statistical significance among baseline conditions.