Abstract
To foveate a visual target, subjects usually execute a primary hypometric saccade (S1) bringing the target in perifoveal vision followed by a corrective saccade (S2). However, in some subjects, saccade behaviour is characterized by the execution of more than one S2. Are these S2 pre-programmed or do they only depend on post-saccadic retinal error? To answer this question, we examined the saccade behaviour of 6 subjects by varying the timing of visual target presentation and by modifying the visual background. Subjects were instructed to perform saccades towards one out of three visual targets. In 20% of the trials, the target was slightly displaced at S1 onset (so-called double step paradigm, used to force the saccadic system to unconsciously integrate a new target position) and was maintained until the end of S1 (‘Brief’), until the start of the first S2 (‘Medium’), or until the last S2 completion (‘Long’). In the remaining 80% trials, the target location was unchanged (‘Fixed’). Experiments took place either in a complete dark room (Dark condition) or in a dimly lit room with an added random dots background (Background condition). Detailed saccadic parameters of S1 and S2 were analysed offline. Results showed that S2 performance was less accurate for Brief and Medium targets conditions, suggesting that the duration of visual integration is a main factor responsible for corrective saccade accuracy. Surprisingly, although S2 accuracy decreased in the Background condition compared to the Dark condition, S1 accuracy had an opposite effect. Finally, the Fixed target condition showed a remarkable effect: as compared to the Dark condition, the Background condition decreased saccade latencies but kept the same accuracy, and the number of S2 decreased in hypometric subjects. The role of attention and the post-saccadic processing of visual information are discussed.