Abstract
When viewing durations are as brief as 100ms, humans can estimate the distance to floor-level targets with great sensitivity (slopes near 1) and a slight bias toward underestimation. Such bias disappears as glimpses are extended, presumably affording eye movements to successfully extract and sequentially integrate local ground information. Recent eye tracking data has revealed no effect of eye movements on one’s ability to make distance judgments during long viewing durations, however. We conducted two experiments to confirm a crucial role for surface information that presumably must be extracted even in the absence of eye movements. All subjects were exposed to both an occluded and a full-viewing condition with block order manipulated. Subjects were instructed to hold gaze on the object once fixated for the duration of the trial (5 seconds) and judged distance via blindwalking to targets 3-6 m distant. In Experiment 1, the occluder obstructed ground surface between observer and target. There was a modest effect of occlusion on sensitivity but only when the occluded condition preceded the full-viewing condition (slope = 0.85 versus 1.01). Indoor environments contain other surfaces (walls and ceiling) which may bias distance judgments when local ground patches are obscured. In Experiment 2, participants viewed the room through a pair of goggles with a small (~ 15[sup]o[/sup]) square aperture. Such a restricted field of view profoundly limited potentially informative cues, leading to lower sensitivity and greater bias towards underestimation when the aperture condition preceded the full-viewing condition (slope= 0.74 versus 1.09). There was no effect on sensitivity when the full-viewing condition preceded occlusion. The results suggest that the nearby ground plane is not as crucial in indoor environments, presumably because other surfaces provide similar reference frames. Additionally, these sources of information need not be extracted with eye movements necessarily– covert deployments of attention are sufficient.
Meeting abstract presented at VSS 2013