Abstract
To perform an anti-saccade, participants need to inhibit the saccade in the direction of a peripheral target, to program and execute a saccade toward the target mirror position. The variables of interest are the percentage of saccade in the wrong direction (errors) and the antisaccade reaction time. As review by Everling and Fischer (1998), a large number of clinical studies have been conducted with anti-saccade tasks. The task is a reliable and a sensitive measure often used to dissociate the different components of saccade preparation and execution. Therefore, anti-saccades are proposed as a diagnostic tool to investigate different pathologies like schizophrenia, Parkinson or mood disorder. This study proposed to investigate: (1) the reliablity of anti-saccade tasks (2) the effect of repeating several sessions of anti-saccades and (3) the interaction with other oculomotor task. In fact different studies dealt with contradictory results (for a review see Smyrnis, 2008). We asked 32 participants to perform anti-saccades. For 16 participants, the anti-saccades were repeated in all trials (block paradigm). For the other participants the anti-saccades were interleaved with prosaccades and nogo trials (mixed paradigm). Each participant repeated 3 times the experiment (sessions 1, 2 and 3) with one week beetween. We found that anti-saccade tasks in both paradigms are reliable (Intra-class correlation and Pearson correlation coefficients). We found a significant effect of the session on saccadic reaction times and saccade errors for both paradigms. This effect is mainly due to a large decrease of reaction times and errors between the two first sessions. The decrease is not significant between the two last sessions. To conclude, when asking participants to perform anti-saccades in mixed or block paradims measures are reliable. However, one should be conscious that there is a possible learning effect that could interact with any clinical prescription.