Abstract
When two synchronized moving gratings are separated by a gap, and oriented along the axis connecting the gratings, faint bars are perceived in the gap (Tynan & Sekuler, 1975). One recent functional magnetic resonance imaging (fMRI) study reported that this visual phantom illusion emerges in early visual areas (i.e., V1 and V2) (Meng et al., 2005). However, it is possible that the visual phantom illusion emerges in higher visual areas (undetected by this recent study because of the larger point-spread function of these higher areas) with feedback as a potential explanation for the activation found in V1and V2. Here we used a simple psychophysical method to test this hypothesis. We predicted that if the visual phantom illusion originates in V1/V2, then no phantom should be perceived when one grating is presented in the left visual field and one in the right visual field. Our reasoning is as follows. First, note that in the absence of visual feedback, neither V1 representation can “know” about more than one of the gratings, because i) visual areas V1 and V2 in each hemisphere get bottom-up input only from the contralateral visual hemifield, and ii) no direct connections exist between the two hemispheres in V1 and V2, except for visual information presented approximately 2–3° around the vertical meridian. Second, a single grating is not sufficient to produce the illusion. We found that visual phantoms do, in fact, occur across hemifields. Furthermore, the across-hemifield phantoms are equally strong, and exhibit the same properties as documented in the literature. Thus, we conclude that the visual phantom illusion emerges in “higher” visual areas, and not V1. These findings suggest a critical role of the top-down processes in the constructive nature of visual perception.