Abstract
Purpose. Visual discrimination in the periphery is limited by crowding, and previous research suggests that crowding effects vary across the visual field. Specifically, the upper visual field is more crowded (less sensitive) than the lower visual field and the vertical meridian is more crowded than the horizontal. To distinguish the contributions of perceptual and decision factors to crowding in natural images, we applied Luce’s Choice Model to describe sensitivity and response bias across the visual periphery. Method. In a spatial 4AFC task, observers viewed static natural images and identified the location of a target patch of “dead leaves” (overlaid ellipses of random size, orientation and aspect ratio), superimposed at one of four directions relative to fixation (N, S, E or W) and one of three eccentricities (2, 4 or 8°, varied across blocks). A staircase converging on the level of 50% correct controlled the size of the target patch. The level of crowding at each location was assessed by the patch size corresponding to threshold performance. Results. A conventional analysis that ignored bias effects yielded several characteristic crowding asymmetries: for two of three observers, the upper visual field was less sensitive than the lower, and the vertical meridians were less sensitive than the horizontal. When performance was considered as a function of both sensitivity and bias, sensitivity decreased linearly with eccentricity (a crowding signature), but only the vertical-horizontal asymmetry remained consistent. The previously inferred anisotropy was thus better described as a difference in response bias. That response bias changed across directions but not eccentricity suggests possible contribution of a motor bias. Conclusion. These results highlight the importance of considering the dual contributions of sensitivity and response bias to performance accuracy in multiple alternative forced-choice paradigms. Failure to account for response bias can lead to erroneous conclusions concerning anisotropic sensitivities.