For and Against Spatial Coding of Saccades

Are target locations coded in retinotopic or in spatial coordinates? Recent neurophysiological experiments supporting these opposite points of view are reviewed. We start with the classical Hallett and Lightstone's double-step paradigm (1976) which established that, in the absence of other cues, an internal representation of eye position is available to locate flashed targets. However, this internal signal is quite distorted at the time of saccades and thus is responsible for considerable mislocalizations. We then look at the evidence for spatial (absolute) coding in single-unit recordings from various brain structures and conclude that this approach has not yet been successful in arriving at a consensus of opinions on the mode of coding and the place where it is effected in the brain. Studies using microstimulation to probe the role of oculomotor structures (colliding saccade paradigm) suggest that the egocentric anchoring of maps occurs in the brain stem and that such maps deal with objects and are essentially allocentric.