Antisaccades exhibit diminished online control relative to prosaccades.

Convergent evidence suggests that stimulus-driven saccades (i.e., prosaccades) are mediated via online trajectory modifications (e.g., Gaveau et al. 2003). The goal of the present investigation was to determine whether manipulating the cognitive demands of a saccade influences the extent to which the response's trajectory is structured online. To that end, participants completed pro- and antisaccades (i.e., 180 degrees mirror-symmetrical transformation) to target stimuli that were continuously visible (Experiment 1) or occluded (Experiment 2) during the response. To index trajectory modifications, we computed the proportion of variance (R (2) values) explained by the spatial location of the eye at 10% increments of normalized movement time [i.e., 10, 20, ... 80, 90% of movement time (MT)] relative to the saccade's ultimate movement endpoint. The basis for this analysis is that between-task differences in the magnitude of R (2) values reflect differences in the use of feedback for online trajectory amendments. Results indicated that antisaccades produced larger R (2) values (from 40 to 80% of MT) as well as less accurate and more variable endpoints than their prosaccade counterparts. Such a pattern of results indicates that antisaccades were not controlled online to the same degree as prosaccades. In particular, we propose that the cognitive nature of the antisaccade task disrupts the normally online operation of saccade networks and renders a mode of control that is not optimized for feedback-based trajectory amendments.