Effects of training on memory-guided saccade performance.

Sensory-motor transformations are often studied using memory-guided movements to small numbers of targets. Whether target locations are directly converted into motor plans on every trial, or subjects use targets to select one of a small number of previously memorized trajectories is unknown. Well-trained monkeys made memory-guided saccades to familiar or nearby novel targets. Performance was superficially similar under the two conditions. However, saccades to novel targets close to the vertical meridian were repulsed away from the nearest familiar target. These findings suggest that sensory-to-motor transformations are performed on every trial, but that previous experience may bias the transformation.