Purchase this article with an account.
Brent Carpenter, Cheryl Olman, Daniel Kersten; Detection of unusual shadows is faster in scenes with weaker 3D cues. Journal of Vision 2015;15(12):1084. doi: https://doi.org/10.1167/15.12.1084.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In the process of ascertaining the properties of objects, the illuminant is often discounted and shadow information is lost. When searching for oddly angled shadows in search arrays, observers are faster for inverted displays than for upright displays (Rensink and Cavanagh, 2004). Faster search in inverted displays is presumably because observers do not have to use attention to retrieve discarded shadow information, since the scene inversion interrupts the usual process of discounting shadows. However, previous studies have typically used simplified stimuli with flat shadow-casting objects presented against a background surface lacking strong depth cues. The present study compared search times under conditions in scenes with rich vs. impoverished 3D cues, and with right-side up vs. inverted presentation. The 3D-rich condition provided linear perspective cues to depth, realistic shadows, and 3D-shaded casting objects. The impoverished scenes lacked shading, the shadow casting objects appeared flat, and the only cues to depth were caster image size and cast shadows. Search arrays contained 2, 6 or 10 casters. On 50% of the trials one shadow was cast at an angle 30 degrees away from the illuminant direction indicated by the rest of the lighting cues in the scene. Detection times with impoverished scenes were significantly faster. Further, when all surfaces lacked shading, observers' search efficiencies were independent of set size. Inversion of the search arrays did not affect observer's time to find the distractor in either the rich or impoverished 3D scenes. Our results suggest that the level of 3D detail in a scene has a larger impact on search times than the global manipulation of inversion.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only