Abstract
Saccadic eye-movements relocate visual input in the primary visual cortex (V1). Using fMRI, we demonstrate pre-saccadic information is transferred to new positions in V1 to support post-saccadic processing. Participants (n=6, two sessions each) viewed a face or house stimulus in the left peripheral visual field, which is processed by right hemisphere V1. Participants were then cued to make a saccade across the stimulus, bringing the post-saccadic image into the right visual field, which is processed in left hemisphere V1. During the saccade, contingent on saccade onset, the stimulus either 1) remained the same, 2) changed, or 3) disappeared. As expected, in the retinotopic position corresponding to the post-saccadic stimulus, we found above-chance multi-voxel decoding for images that remained the same. However, when the stimulus changed during the saccade, decoding was reduced, yet still above chance. Finally, when the stimulus disappeared during the saccade, it could not be decoded at the new retinotopic position. The difference in decoding between the same and change trials suggests stimulus information encoded prior to the saccade is transferred across hemispheres. Information transference facilitated post-saccadic processing when the stimulus remained the same. When the stimulus changed, decoding accuracy decreased suggesting interference between transferred pre-saccadic information and the processing of post-saccadic input. The inability to decode in the disappearance condition suggests a post-saccadic reference frame may be necessary for pre-saccadic information to be retinotopically relocated within V1. Our results suggest that stimulus feature information is transferred to new retinotopic positions in V1 with each saccade. Importantly, this information is likely fed-back to V1 from higher cortical regions, since there are no direct anatomical connections between homotopically-aligned portions of right and left hemisphere V1.