Abstract
There has been an ongoing debate as to whether optic flow is sufficient to judge heading or whether extra-retinal information is required. A more recent debate has been whether human locomotion can be controlled purely on the basis of egocentric visual direction without reference to optic flow (Rushton et al, 1998 ; Warren et al, 2001). We created a driving simulation where retinal flow (RF) and visual direction (VD) information could be manipulated independent of one another. Observers sat in a projection booth with a 90 by 64deg FoV, with a simulation of locomotion at 8m/s across a textured ground plane, and used a force-feedback wheel to steer toward target gates. We systematically manipulated the information from optic flow or other directional cues in a balanced design so we could predict over-steer or under-steer if that particular cue was being used. The results supported the role of both RF and VD information in controlling steering with significant under/over-steer in the predicted direction when either information source was biased (p <.01). The weighting attached to either source of information appeared to be approximately equal for the conditions we created and reinforces the argument of Warren et al (2001), that if retinal flow is salient, it will be used in active locomotion. A further experiment manipulated extra-retinal (ER) information and supported a model where all 3 sources (RF, VD, ER) are combined in an additive manner to provide robust locomotor control. Research supported by the UK EPSRC GR/R14644.