Abstract
Anti-saccade are eye movements directed in the opposite direction of a visual target. Because the visual target and the saccade goal are decoupled, it has been suggested that the competition between those two locations need to be resolved for the execution of a correct anti-saccade. However, it remains unclear how this competition is resolved temporally and spatially. To examine how anti-saccade metrics reflect competition across time and space, we tested 14 participants on a pro-saccade paradigm as well as three different spatial configurations of anti-saccades: 90° away across hemifields, 90° away within the same hemifield, and 180° away with different target eccentricities (4, 5.5 and 7 degrees). We measured anti-saccade metrics including error rates, amplitudes and saccade endpoints and reaction times. We observed a dynamic reduction of error rates as a function of saccade reaction times. Moreover, compared to pro-saccade endpoints, we found a spatial bias toward the visual target in the endpoints of anti-saccades across all anti-saccade paradigms. This bias was specific to the vector inversion direction of each anti-saccade paradigm. Further, the magnitude of the bias was modulated by the different eccentricities tested. Taken together, these results show evidence for competition between the visual target and saccade goal location that varies dynamically across time and space during anti-saccades.