Humans rely on predictive control to maintain stability while walking in dynamic environments, yet the strategies driving this control remain unclear when environments are inconsistent or uncertain. To investigate this, healthy adults performed repeated, goal-directed stepping trials while a robotic device applied lateral force field disruptions to the pelvis in alternating directions (left or right). We manipulated the force field's Consistency (how often the force direction switched from trial to trial) and Certainty (how predictable the switches were). We quantified motor adaptation (error reduction) and sought evidence for two predictive strategies: pattern prediction (detecting a global trial pattern) and carryover prediction (assuming the next trial matches the previous). Adaptation profiles revealed a surprising finding: Consistency, not certainty, drove predictive control. When trials were Consistent, participants pursued appropriate predictive strategies (pattern or carryover). However, when trials were Inconsistent, participants exhibited counterproductive strategies, such as carryover prediction (despite the next trial most often being opposite) or no prediction (despite full certainty of each forthcoming trial). Error reduction was likewise dominated by consistency, with certainty exerting negligible influence. These findings, counter to long-held observations in upper-limb control, suggest the central nervous system faces unique challenges in whole-body control of walking and cannot rely on the simpler adaptation strategies observed in isolated limb control. This raises new, important questions for control of walking and whole-body motor adaptation.
Grover et al. (Mon,) studied this question.