Indications of extent depend very much upon how the observer is
asked to respond. When observers attempt to adjust the magnitude of an
extent lying in depth to match that of an extent lying in a frontoparallel
plane, the results suggest a strong foreshortening in perceived depth at
intermediate to far viewing distances. Loomis, Da Silva, Fujita and Fukusima
(1992), however, found that when extents are indicated by blindfolded walking,
there is no evidence of this foreshortening. Four hypotheses that might
explain these apparently paradoxical results are discussed: (1) Both indicators
are controlled by perception, but cognitive processing (possibly unconscious)
influences the two indicators to a different extent; (2) There are distinct
spatial representations underlying the walking indicator and the matching
indicator; (3) Different stimulus information is processed by the visual
system depending on which indicator is to be used, but the processes underlying
both indicators are otherwise the same; (4) Perceived extents are not completely
specified by the perceived locations of the endpoints specifying the extents;
the matching indicator is more responsive to perceived extents, whereas
the walking indicator is more responsive to perceived locations.
Six experiments are reported that investigated the reasons underlying the apparently paradoxical results in Loomis et al. (1992). The experiments were conducted under well-lit viewing conditions. Under objective instructions, observers indicated the magnitude of depth and frontal extents, as well as egocentric distances, by a variety of methods. Walked indications of extents were rarely foreshortened in depth, although in some circumstances they did show large and robust constant errors. By contrast, verbal and matching indications of extents were often foreshortened in depth to some degree. Only Hypothesis 1 above was consistent with all the results, suggesting that the differences in matching and blind walking indications of extents are due to differences in the cognitive processing that intervenes between perception and behavior. Though the data presented here did support Hypothesis 4, this hypothesis may still play a role in a more complete explanation of the observations reported here and elsewhere.