Abstract
Previous research has established that humans can walk without vision to previewed targets without large systematic error, and that they can do so with equal accuracy under brief (150 ms) and extended ([[gt]]5 sec) viewing conditions. The present study determined the duration of viewing required to support accurate blindwalking performance, and began investigating the availability of distance and depth cues at the earliest stages of viewing. Participants binocularly viewed targets through a liquid crystal shutter that provided precisely-controlled glimpses of a real, naturally-lit indoor room. Glimpse durations were 50, 100, or 200 ms, after which there was either a visual mask or no mask. In Experiment 1, participants walked to a cone placed on the floor, 3–5 m distant. Performance was relatively good and equivalent at all three exposure durations (−10% error) but declined modestly in the 50 ms viewing condition when the masking stimulus was imposed. The results indicate that the cues required to support good blindwalking performance are available at or near the shortest manipulated time frame. In Experiment 2, the cues of relative size and angular declination were removed by suspending ball targets at eye level and by holding the angular target size constant. Performance was generally poorer (−20% error) than Experiment 1, and walked distance was not affected by target distance in the 50 ms viewing condition when the masking stimulus was imposed. However, performance improved systematically with increased exposure duration, suggesting that, unlike in Experiment 1, the cues provided by this context were not yet fully available at the shortest manipulated time frame. The study not only indicates a lower limit on the duration of viewing needed to support good blindwalking performance, but also suggests that different sources of information about distance and depth may take longer to extract.