<p>Sustained attention is often treated as a single, uniform construct, largely interchangeable with vigilance. We propose that sustained attention instead comprises two distinct yet complementary modes of engagement: an active mode, characterized by the need to process continuously changing task information, and a passive mode, defined by monitoring readiness under uncertainty. To test this hypothesis, participants completed two versions of a Multiple Object Tracking (MOT) task. In the Active-MOT (A-MOT), participants continuously tracked moving targets, and in the Passive-MOT (P-MOT), they relied on sustained monitoring until target identities were revealed (unpredictably) late in each trial. Behavioral performance and pupil diameter were recorded as proxies for engagement dynamics and arousal regulation, revealing clear dissociations between the two tasks. Behaviorally, the A-MOT was characterized by lower accuracy and greater sensitivity to task difficulty than P-MOT. These differences were likely driven by attentional engagement: Whereas the A-MOT was accompanied by sustained, load-dependent pupil dilation reflecting continuous cognitive effort, P-MOT yielded smaller phasic pupils punctuated by transient, load-modulated dilations at target assignment, signatures of reactive, event-driven engagement. Pretrial pupil size was larger in P-MOT, indicating heightened anticipatory arousal under uncertainty, but it predicted performance only in A-MOT, suggesting distinct preparatory engagement states. Together, these findings demonstrate that sustained attention operates through two separable modes governed by distinct arousal-control dynamics. Recognizing this dual-mode architecture reframes sustained attention not as a singular capacity to remain alert but as the flexible coordination of processing evolving task information and monitoring readiness across time and context.</p>