Across four experiments, we manipulated features of a simple reaction time (RT) task to examine the effects of such features on vigilance. In Experiment 1, we created simple reaction time “game” that pitted participants against two computerized avatars. In one condition, participants were awarded points, while the other condition did not receive points. Performance in the two conditions did not differ, but both conditions showed faster RTs and shallower time-on-task performance decrements compared to a standard psychomotor vigilance task. In Experiment 2, we removed the competitive feature but retained the point system. In this case, participants without a point system showed a steeper performance decrement than those with a point system. Experiments 3 and 4 replicated these effects and corroborated their findings with pupillometry. Participants in both conditions of Experiment 3 (competitive task) and the points condition of Experiment 4 showed larger task-evoked pupillary responses than participants in the no-points condition of Experiment 4. These findings challenge the notion that time-on-task performance decrements are caused by resource depletion (Smit et al., 2004), and are better explained by motivational control (Hockey, 2011) or cost-benefit theories (Boksem & Tops, 2008; Kurzban et al., 2013) of mental effort and cognitive fatigue.