Abstract
A recent study (Williams et al., 2008) reported the surprising finding that the region of cortex representing the fovea contains information about objects presented in a peripheral retinal location, far from the fovea. Although other cases of feedback to retinotopic cortex have been reported before, this case is unusual in that the feedback apparently constructs a totally new representation in a different cortical region from the feedforward representation. Why is this new feedback representation constructed in foveal cortex? One possibility is that foveal cortex is engaged because it is a unique piece of neural hardware ideally suited for fine-grained visual computations (Mumford, 1991). Here we tested this hypothesis in individuals with macular degeneration (MD) who have lost all foveal vision, and who use a “preferred retinal locus” (PRL), a spared part of the peripheral retina, for face recognition, reading, etc. – essentially the functional equivalent of a fovea. If feedback information arises in foveal cortex because of its unique physical/computational properties, we predict the peripheral object information will be found in the foveal cortex in MD individuals, even though they have lost all foveal vision. By contrast, if feedback information arises in foveal cortex for some other reason (e.g., this is the location where most detail visual processing goes on in this person), then we predict the peripheral object information will be found in the region of cortex corresponding to the PRL, not the fovea, in MD individuals. Using fMRI, we found that the region of cortex representing the PRL, not the fovea, contained information about objects presented in another peripheral location. This result indicates that feedback is sent to foveal cortex in normally-sighted subjects not because of any unique physical characteristic of foveal cortex per se, but because this region of cortex is used most intensively for detailed visual processing.
This work was supported by NIH grant EY13455 to NK.