Vestibular, podokinetic and visual information can all contribute to the sense of orientation as we walk around. Each provides the brain with different qualitative information in different reference frames. How are they fused to create a coherent internal representation of self-motion relative to the external environment? How this is used to guide locomotion? Each of these systems can be stimulated independently to evoke signals of rotation about the same vertical axis; the podokinetic on a rotating platform, the visual with optokinetic rotational flow, and the vestibular with electrical currents and appropriate head posture. Stepping trajectories during and after conditioning stimuli reflect how the three signals, or channels, are fused to generate a unitary perception. With rotation about the vertical, balance responses are left out of the behaviour so that just the orientation behaviour is observed. Conditioning of individual channels shows channel specific adaptations but expressed according to the calibrated states of the other channels. For example, a vestibular signal of rotation in the presence of a stationary visual signal is rapidly recalibrated to the visual signal. When vision and the applied vestibular stimuli are removed, the after-effect reveals the vestibular adaptation and how it had been recalibrated to vision during conditioning. The data are consistent with a conceptual dynamic forward model that incorporates adaptation within individual channels, recalibration between channels of adapted signals, and reweighting between channels according to the task and “belief”.