Kent State University  - Sun Microsystems Inc.

Familiar Suspended Objects Appear Smaller Than Actual Independent Of Viewing Distance

Michael K. McBeath, John G. Neuhoff, Diane J. Schiano

Abstract

In the present study, observers of basketball hoops and traffic light exhibited a robust tendency to underestimate actual size by about 30%, independent of viewing distance.  When observers judged the size of these familliar, suspended, 3-D objects, they  displayed accurate size constancy, but with a constant level of underestimation.

Numerous past studies confirm that observers may systematically misestimate object size as a function of viewing distance when object familiarity or depth information are reduced.  Yet size estimations for familiar work suggests that many perceptual illusions in 2-D pictures and drawings are due to impoverished depth information that does not allow an accurate assessment of implicit depth.  An applied implication is that size  distortion in simulated 3-D computer displays may also result from lack of depth realism.  In the present work, we tested accuracy of size judgements under conditions where observers seem to have an accurate representation of viewing distance.

In our experiments naive observers stood at varying distances from either a basketball hoop or a traffic light.  In the basketball setting, observers indicated perceived size of the hoop relative to the size of a basketball at viewing distances ranging from 110 to 50 feet.  In the traffic light setting, observers compared perceived size  of a 12 inch traffic light to a set of circles drawn on a posterboard held several feet away.  Here, viewing distances ranged from 15 to 500 feet.  Findinngs in both settings indicated no significant underestimation in size at all viewing distances.  Thus, in both settings viewers maintained size constancy over viewing distance, but with a constant level of size underestimation.

If observers maintain size constancy over viewing distance, it follows that they have an accurate spatial representation of distance. Successful basketball shooting also implies an error free proprioceptive representation of space.  An accurate spatial representation suggests that factors other than misestimation of viewing distance are principal causes of the observed size illusion.  It also follows that size distortion of suspended objects in pictures and simulated 3-D computer displays would not be eliminated with added 3-D realism.

The findings confirm the occurrence of a robust three-dimensional illusion under information-rich, real world viewing conditions in which observers seem to display an accurate spatial representation of  distance.  Further research is planned to assess parameters that contribute to this illusion in both real world and virtual reality settings.