Abstract
Judgments of the orientations of 2D lines show systematic biases that can be described as exaggerating deviations from horizontal. These biases are not due to numeric estimation biases. An orientation of about 37° from horizontal is judged to be equidistant from horizontal and vertical. The same orientation is judged to be 45° from horizontal and 45° from vertical. One hypothesis concerning this distortion is an experience-based anisotropy in the distribution of orientation sensitive cells in visual cortex. An alternative view is that these biases are intrinsic to the encoding of spatial orientation. To tease apart these two possibilities we oriented observers at 45° from vertical and asked them to do two kinds of tasks. One task was a psychometric measurement of perceived vertical or horizontal using a forced choice method. Judgments of visual stimuli, presented in a circular aperture that removed any external visual references, were made either with respect to an egocentric (bodily) or exocentric (gravitational) reference frame. Over 150 participants were tested. Both within and between subjects, variability was much less for judgments made with respect to the exocentric (perceived gravitational) reference frame. Similarly, when required to make numeric estimates of orientation with respect to the two types of reference frame, estimates with respect to perceived gravitational frames exhibited the standard bias exaggerating deviations from horizontal. Estimates with respect to bodily frames did not. These findings are consistent with the precedence of gravitational reference frames in visual experience and imply that 2D orientation asymmetries between horizontal and vertical are not caused simply by biases in the distribution of early cortical orientation analyzers. Precision was greatest when observers were upright so that gravitational and bodily vertical were aligned, but for tilted observers, judgments made with respect to gravitational reference frames were most similar to judgments made upright.
Meeting abstract presented at VSS 2012