Error-correcting mechanisms in large saccades

Abstract The initial saccade toward a very peripheral stimulus is generally hypometric and necessitates a corrective saccade. In a first experiment the stimulus was cut off at the onset of the initial saccade. If the hypometry was important a secondary saccade occurred reducing the error, but not truly corrective. If the hypometry was normal (10%), almost no secondary saccade occurred. In a second experiment the stimulus was cut of during the deceleration phase of the initial saccade. Secondary saccades (which were truly corrective) occurred very often, provided the velocity of the eye at the time of cut off was inferior to about 100°/s. It is suggested that extraretinal signals for error detection are not very sensitive, and that in normal conditions reafferent visual feedback can be taken into account before the end of the saccade, to generate a secondary saccade encoded in absolute position, though the information is still available in retinal position.