Abstract
Visually-guided behavior requires information about spatiotopic (world-centered) locations, but initial visual input is retintopic (eye-centered). Does spatiotopic information reside in higher-level, spatiotopically-organized visual maps, or is it continually reconstructed from retinotopic information with each eye movement? Recent work favors the latter option; to a surprising extent, visual processing occurs in native retinotopic space, even in higher-level visual areas (Golomb, Chun, & Mazer, 2008 J.Neurosci.; Golomb & Kanwisher, VSS 2010). These findings suggest a counterintuitive hypothesis: despite our conscious experience of an effortlessly stable spatiotopic world and our lifetime of practice with spatiotopic tasks, human behavior might actually be better in retinotopic coordinates than in the more ecologically relevant spatiotopic coordinates. To test this idea, we had subjects perform two sessions of a spatial working memory task: one in which they memorized the spatiotopic (absolute) location of a cue, and another in which they memorized the retinotopic location (relative to the eyes). After a delay – which included 0, 1, or 2 guided saccades – subjects reported the memorized location by moving the mouse to the appropriate place on the screen. The error in distance between their reported location and the true location was taken as a measure of the precision of spatial memory. On trials with no saccades, spatial memory was equally precise during the retinotopic and spatiotopic tasks. However, after an eye movement, subjects were significantly more precise at remembering retinotopic locations than spatiotopic locations. This difference grew with each eye movement, such that spatiotopic memory continued to deteriorate, while retinotopic memory remained largely intact. Thus, while we are capable of spatiotopic memory, it is not as reliable as retinotopic memory, and the loss in fidelity with each update is not an inevitable cost of eye movements, but rather a consequence of converting the information from its native retinotopic coordinates.
NIH R01-EY13455 (N.K.) F32-EY020157 (J.G.).