Egocentric distance judgments are impaired when the field of view is limited. A prominent account suggests that the impairment derives from an inadequate representation of the geographical slant of the ground surface, which depends on the sequential integration of ground cues. In support of this view, near-to-far scanning with a restricted view has been shown beneficial in an outdoor experimental setting. In contrast, performance in an indoor setting has shown that occluding the nearby ground plane results in smaller deficits than occluding the greater room space. The present study aimed to reconcile these outcomes by matching the designs and manipulations used in indoor and outdoor settings. We report three experiments that compare blind walking performance with a restricted view (≈15° square aperture) to that with an unrestricted view of the target and scene context. Restricted viewing conditions were near-to-far scanning, far-to-near scanning, and/or with head held steady. In Experiment 1 (indoors), errors were greater with the restricted view (-25%) than with an unrestricted view (-15%) regardless of viewing condition (steady or near-to-far scanning) and regardless of block order. In Experiment 2 (outdoors), which employed a completely randomized design, target distances were actually overestimated with the restricted view (+8%) and nearly unbiased with an unobstructed view (-2%) regardless of viewing condition (near-to-far or far-to-near scanning). In Experiment 3 (outdoors), there were no reliable differences between viewing conditions (restricted steady vs. unrestricted view) regardless of block order. The pattern of results indicates strikingly different biases in two different but similarly reduced-cue contexts. Further, the observed bias even with an unrestricted view differed between indoor and outdoor contexts. Aspects of these data are not predicted by previous work and suggest the need for a contextual-scaling framework. A breakdown of context effects and an analysis of potential sources for these will be discussed.
Meeting abstract presented at VSS 2014