Retinal flow controls gait during natural locomotion

It has long been thought that self-generated flow patterns play a role in locomotion and are used to guide steering towards a goal. However, lateral oscillations of the head through the gait cycle lead to continuous variability in the momentary direction of heading, making it unsuitable for steering (Matthis et al, 2022). While the information in retinal flow patterns is not suited for steering, it has been shown to influence postural control. In this experiment we explore whether subjects have an internal representation of expected flow patterns generated from body movement and whether deviations from these patterns influence gait parameters. We adjusted how measured head motion modulated the resulting retinal flow patterns. In particular, we increased or decreased the gain of the head motion in either the forward or lateral direction so that the resulting image motion was either increased or decreased relative to the subject’s actual motion. We measured subjects’ feet, hip, torso, and head in 6 degrees of freedom tracking space while subjects walked a 5-8m path in an unconstrained goal-directed walking task in a virtual environment. Subjects reliably changed their step length, step duration, and step width in response to the gain manipulations. Forward gain modulation of +/-22.5% decreased step time by 4-16% between low and high gain conditions. Lateral gain +22.5% increased step width by 33-60%, while lateral gain -22.5% decreased step width by 40-53%. These results suggest that pedestrians have an internal representation of the retinal flow generated by their natural gait, and that retinal flow is used to continuously monitor gait during natural locomotion.