Abstract
Vision requires a reference frame. To what extent does this reference frame depend on the structure of visual input, not just local retinal coordinates? This question is especially relevant for perception of dynamic scenes, where the reference frame defined by visual input may move relative to the retina. To answer this question we used novel modifications of two well-know paradigms: biological motion and masked pentagon (Lorenceau & Shiffrar, 1992). These paradigms were chosen because a simple manipulation (inversion and absence of visible apertures) transforms a perceptually structured stimulus into a set of disorganized elements. Biological motion displays were created by placing Gabor patches (sigma = 5 arcmin; 4 cycles/deg) on the major joints of a human walking in place. Gabor patches oscillated (2 Hz) either coherently or with some phase difference. In a spatial 2AFC task, observers discriminated motion coherence of Gabor patches defining either upright or inverted biological motion. In an analogous biological motion experiment, observers discriminated temporal coherence of counterphasing black/white disks. In masked pentagon displays, the pentagon was presented behind five apertures and translated along a circular path. Each pentagon side was defined by a grating oriented parallel to the side. Gratings oscillated either coherently or with some phase difference within the limits of the pentagon borders. The pentagon apertures were either visible or invisible. In a temporal 2AFC task, observers identified the interval in which five gratings oscillated coherently. Across all displays and tasks, coherence discriminations were more accurate when the stimulus was perceptually structured (upright biological motion and pentagon behind visible apertures). Evidently, information is extracted more efficiently from a perceptually organized reference frame. The changing structure of visual input can provide a reliable non-retinal reference frame for vision.
Supported by EY07760 and P30-EY08126.