Canonical perspectives of things were said to be easier to comprehend than noncanonical viewpoints, according to earlier research. All of these investigations, however, required participants to name the item, which is, at best, a cognitive process and at worst a late perceptual process. 

By eliminating the explicit requirement to identify the objects, researchers extended the technique to early vision. They specifically challenged participants to distinguish between intact and scrambled versions of a set of rapidly presented items when they were seen from either normal or unusual perspectives. Participants were simply required to distinguish the item from noise, not name it (scrambled). 

When items were presented in normal orientations as opposed to unusual depth rotations, participants were more adept at differentiating between them (Experiment 1). The same impact, however, did not approach statistical significance (Experiments 2 and 3) for objects shown in atypical picture plane rotations (as opposed to typical ones), indicating that certain informative perspectives may be crucial to this effect. 

They interpreted improved perceptibility as resulting from the strong real-world statistical regularity of both these objects and good exemplars and likely situations.

Reference: jov.arvojournals.org/article.aspx?articleid=2783815