Abstract
McBeath, et. al. (1992) found that observers underestimate the size of traffic lights across a wide range of viewing distances. The present study further investigated this failure of size constancy. Three groups of participants viewed a traffic light from distances of 20, 60, and 120 m, respectively. Each participant estimated the size of the light's 30.5 cm (diameter) lenses by selecting, from a set of nearby comparison circles, a circle that matched the lenses' size. A fourth group estimated the size of traffic-light lenses from memory with no traffic light visible. The mean size estimates made by all four groups were approximately 30% smaller (in diameter) than the lenses' actual size. There were no significant differences between the four groups' size estimates.
After making their size estimates, participants in the fourth group viewed a traffic light from a distance of less than 2 m, and were asked to rate the light's size relative to what they expected and to rate their surprise regarding the light's size. The majority of participants rated the light as “much larger” than expected and reported that they were “very surprised,” giving maximum ratings on ordinal scales for these responses.
The results indicate that perception of traffic-light size is not veridical at viewing distances of 20 meters and greater. The finding that estimated size does not depend on viewing distance or on the presence of a visible traffic light further suggests that observers rely on assumed size when estimating the size of a distant traffic light. For most observers, the light's assumed size does not match its actual size and size constancy is not achieved. These results are consistent with the hypothesis that size constancy for distant objects is not a feature of perception, and that research participants' estimates of distant objects' sizes are based on cognitive strategies in addition to visual information for size.