Human control of upright body posture involves inputs from several senses (visual, vestibular, proprioceptive, somatosensory) and their central interactions. We recently studied visual effects on posture control and their intersensory interactions and found evidence for the existence of an indirect and presumably cognitive mode of interaction, in addition to a direct interaction (we found, e.g., that a 'virtual reality' visual stimulus has a weaker postural effect than a 'real world' scene, because of its illusory character). Here we focus on the presumed cognitive interaction mechanism. We report experiments in healthy subjects and vestibular loss patients. We investigated to what extent a postural response to lateral platform tilt is modulated by tilt of a visual scene in an orthogonal rotational plane (anterior-posterior, a-p, direction). The a-p visual stimulus did not evoke a lateral postural response on its own. But it enhanced the response to the lateral platform tilt (i.e., it increased the evoked body excursion). The effect was related to the velocity of the visual stimulus, showed a threshold at 0.31 degrees /s, and increased monotonically with increasing velocity. These characteristics were similar in normals and patients, but body excursions were larger in patients. In conclusion, the orthogonal stimulus arrangement in our experiments allowed us to selectively assess a cognitive intersensory interaction that upon co-planar stimulation tends to be merged with direct interaction. The observed threshold corresponds to the conscious perceptual detection threshold of the visual motion, which is clearly higher than the visual postural response threshold. This finding is in line with our notion of a cognitive phenomenon. We postulate that the cognitive mechanism in normals interferes with a central visual-vestibular interaction mechanism. This appears to be similar in vestibular loss patients, but patients use less effective somatosensory instead of vestibular anti-gravity mechanisms.