Competition Between Task Accuracy and Habitual Locomotor Behavior Underlies the Speed-Accuracy Trade-Off in Walking

The speed-accuracy trade-off is a hallmark feature of motor behavior, yet its underlying causes in cyclic, whole-body movements such as walking remain to be understood. In discrete, point-to-point movements, reduced accuracy at higher speeds is typically attributed to signal-dependent motor noise and an inability to use online error correction. Walking, however, may be governed by additional objectives, such as minimizing metabolic energy expenditure. To distinguish between these mechanisms, we tested young adults in a precision-walking task requiring them to match target step lengths across a range of treadmill speeds. Accuracy declined systematically as walking speed increased, demonstrating a locomotor speed-accuracy trade-off. Contrary to prevailing theory, removing online visual feedback of foot position did not affect this decline, indicating little contribution from visually guided error correction. Moreover, step-length variability, a task-specific measure of signal-dependent motor noise, decreased rather than increased with speed. Instead, faster walking produced progressively larger biases toward participants' habitual step lengths measured in the absence of targets. Because habitual step lengths are known to closely approximate energetically optimal values, these findings suggest energetic optimization, rather than motor noise or limitations associated with the use of online feedback, as the primary driver of the speed-accuracy trade-off during walking.