Abstract
Visual field inhomogeneities have been the subject of considerable investigation, but some aspects of these inhomogeneities, such as the vertical meridian asymmetry (VMA; Carrasco et al., 2001), are less well understood. The VMA is defined as particularly poor performance on the vertical meridian, directly above the point of regard. Cameron & Rathje (VSS 2006) hypothesized that the VMA depends on stimulus visibility. Here we test this hypothesis. Methods: Using Gabor patches of low (0.5 cpd) and relatively high (8.0 cpd) spatial frequency we concomitantly varied grating contrast and tilt (i.e., the higher the contrast the lower the tilt) in three 2AFC orientation tasks. Stimuli were presented at 1 of 8 locations around an imaginary circle at 4.5 deg. eccentricity and overall performance was about 80% correct. In a second experiment we briefly presented suprathreshold, real-world images to observers in a series of 2AFC categorization tasks (e.g., cat vs. dog) in the same manner as in the first experiment. We expected performance (measured by percent correct, PC) to be more homogeneous for high contrast stimuli. Data were fit with ellipses that were constrained by all but PC at the north location and we calculated visual field inhomogeneity in two ways: (1) the difference between the predicted PC based on the ellipse fit and the observed PC at the "north" location (the VMA) and (2) the ratio of the minor to major ellipse axis lengths, an index of the horizontal-vertical anisotropy (HVA, Carrasco et al., 2001). Results: Performance was homogeneous for real world images and the VMA (and HVA) became less pronounced as contrast increased for both Gabor stimuli. Conclusion: Target contrast/visibility may be a critical factor in the VMA. The lack of inhomogeneity for natural images supports the conclusion that the VMA is a low-level perceptual phenomenon.
Meeting abstract presented at VSS 2013