Abstract
Two very different bias functions may be observed in the study of perceived 2D orientation. One bias function is a symmetrical pull toward vertical and horizontal. This bias has sometimes been argued to be a front-end coding bias due to the overrepresentation of vertical and horizontal in natural scenes and in visual cortex. However, when retinal/cortical and gravitational frameworks are dissociated by rotating observers by 45° the bias is tied to world, not retinal/cortical orientation. Moreover, the bias can be simulated as a range effect when a subset of possible orientations is used for comparison. These observations challenge the dominant view and suggest instead this bias may actually be a comparison or memory bias in which cardinal categories act like implicit range delimiters. The second bias function, which is asymmetrical between vertical and horizontal, seems to be a true perceptual bias of unknown origin: Apparent deviations from horizontal are perceptually exaggerated whereas deviations from vertical are underestimated (e.g., the perceived bisection point between horizontal and vertical is much closer to horizontal). This bias is yoked to the perceived (gravitational) reference frame, but it differentiates between horizontal and vertical. Is this differentiation due to the 3D differentiability of these axes? In 3D space, vertical is a set of parallel vectors (defined by local gravity), whereas horizontal is a set of parallel planes to which gravity is a normal vector. The cognitively distinct status of these two categories may help to explain the asymmetrical bias observed in 2D orientation perception.