Abstract
One of the defining characteristics of peripheral vision is that positional information is largely degraded. In a psychophysical experiment we investigate how positional uncertainty affects appearance within peripheral vision. To this aim we generated spatially disarrayed versions (eidolons) of natural images from the UPenn database. The disarray was determined by a random field confined to a 10° circular area within the image and was characterized by the maximum displacement which could be applied to a pixel (reach). In each trial, observers were exposed to two versions of the same image, the original image and the disarrayed eidolon in a random sequence. Each picture was shown for 1.5 s, interleaved for .5 s by a phase-scrambled image. Observers reported which image appeared disarrayed. Between trials we varied both the reach and the eccentricity of the disarrayed patch. Observers were forced to fixate either the center of the disarrayed patch, which was located left or right in the image, or at locations 10°, 20°, 30° or 40° towards the other side of the image. Accuracy in disarray identification as a function of reach was fitted through psychometric curves, yielding reach thresholds at each eccentricity. The data by six observers showed a clear linear increase in the reach thresholds as a function of eccentricity up to 30°, with average foveal reach thresholds of .143° (std .039) and slopes of .015 (std .005). Disarray detection proved much harder at 40° eccentricity, largely independent of reach. Our results show that features can be displaced to a substantial extent (over .5°) in peripheral vision, before an image loses its perceptual quality of naturalness, in accordance with the hypothesis that in peripheral vision positional information is degraded before it reaches awareness. Eidolons potentially offer a way to visualize the appearance of peripheral stimuli in foveal vision.
Meeting abstract presented at VSS 2016