In the present study we assessed the use of landmarks and scene layout information for the control of locomotion. Observers were presented displays simulating forward locomotion through a random dot field with the horizontal position perturbed by a sum-of-sines function and were asked
to steer and null out the horizontal disturbance of the path of locomotion. The results indicate greater control gain and accuracy when presented with a repeating layout of landmarks as compared to a changing layout of landmarks. Debriefing responses suggest that observers may have implicitly
learned the layout of the repeating pattern. These results suggest that observers use an allocentric representation of the scene for steering control. A model for the control of locomotion is discussed that utilizes both scene-based information and optic flow.