Abstract
Many perceptual variables, like distance, are likely coded logarithmically such that we expect Weber’s law to hold, but angular variables might not share this property. Limits in the precision of angular coding should be distance invariant. A directional error expressed at a far distance will represent a greater Cartesian displacement than at a near distance, but not a larger polar (angular) error. Nonetheless it appears to be subjectively more difficult to perform angular updating for larger distances when walking obliquely with respect to the target. To measure this, we used a facing task. On each trial, participants (N=37) walked blindfolded along a guide wire obliquely to a visually-previewed target until told to stop and were then asked to face toward the target. A compass was used to record facing direction. Similar geometries at scales of 6.6:1 were used for a far-space set-up and a near-space set up with final direction ranging from 55° to 90° with different scales being tested in random order. Variability in produced facing direction (i.e. mean SDs for 8 tested directions) was reliably greater for far targets (19°) than for near targets (13°), t(7) = 3.22, p = .0027. Because variability in perceived walking distance seems to follow Weber’s law (Durgin et al., 2008), the greater angular error for farther distances suggests that the original visual encoding of egocentric distance might not follow Weber’s law. Such a result is consistent with the use of angular variables (i.e., angular declination or gaze declination) to measure egocentric distance if we assume a small constant error in perceived straight ahead added to proportional noise in the coding of angular declination. In this scheme, small errors in the perceived horizon point could produce proportionally greater variability in perceived distance and therefore in angular updating for far targets than near ones.
Meeting abstract presented at VSS 2013