Observers can sight a target 20 m away or more and then walk to it accurately without vision. In contrast to this good performance, this article shows that walked indications of the exocentric separation of two locations exceed the required values by over 70% when vision is obscured. Significantly, these large errors are coupled with a robust lack of depth foreshortening, even under conditions in which visual matches and verbal estimates of extent exhibit strong evidence of depth compression. This article presents evidence that the overshooting errors are due largely to recalibration of locomotor control produced by prolonged exposure to nonvisual walking. The robust lack of depth foreshortening, meanwhile, could reflect a corresponding isotropy in the spatial representation controlling the walking response. More research is needed to confirm this interpretation, however.

No abstract