Abstract
Express saccades are commonly produced under predictable task conditions. Here we report the production of express saccades in a saccade countermanding task, where conditions strongly discourage express saccades. To perform this task, monkeys fixated a central spot whereupon after a variable time the central spot was extinguished and a peripheral target appeared simultaneously at one of two random locations on the left or right (no-stop trials). On ~30% of trials, after a variable delay, the central spot reappeared, instructing monkeys to cancel the planned saccade to the peripheral target (stop-signal trials). Monkeys were rewarded with juice for generating saccades to targets on no-stop trials and for maintaining fixation on stop signal trials. Saccade latencies in this task are notably longer than values obtained in a simple reaction time task because the monkeys adapt to the stop signal by slowing performance. However, we observed some monkeys producing saccades with response times <100 ms. Indeed, we observed bimodal distributions of saccadic latencies, characteristic of express and regular saccades. Simulations revealed that reward rate can be optimized by varying the proportion of express saccades made within a session. The monkeys exploited this opportunity. When production of express saccades was not rewarded, monkeys made significantly fewer. This study highlights the many ways in which cognitive control of visually guided gaze behavior can exploit niches of opportunity.