How can we reconcile this? One possibility is that cognition reflects some putative higher level area with a spatiotopic reference frame that updates the retinotopic representation after each saccade. Arguing against this is the relative paucity of evidence for any explicit spatiotopic representation in the visual cortical hierarchy (Dean and Platt, 2006; Duhamel et al., 1997; Golomb and Kanwisher, 2012). Alternatively, spatiotopic cognition may not rely on an explicit spatiotopic neural representation. It could instead be coded implicitly by a retinotopic representation that is locally updated after a saccade. This would shift a remembered location in the opposite direction of each saccade vector to its new coordinates on the retinotopic map (Pouget and Snyder, 2000). For a midline-crossing saccade, this would entail transfer of the memory trace between the left and right visual hemifields/hemispheres.
