The Timing of Suboptimality: How Temporal Uncertainty Drives Procrastination Behavior

Procrastination, where people engage with an immediately rewarding task when they should work on a task with a delayed payoff, is typically framed as self-regulation failure or lack of motivation. We challenge this view by demonstrating that procrastination emerges naturally from the computational intractability of sequential resource allocation under temporal uncertainty. When time horizons are certain, humans approach mathematically optimal strategies for balancing competing tasks. However, when deadlines are uncertain -- requiring integration over exponentially growing outcome spaces -- participants systematically over-invest in immediate rewards early, then attempt to compensate too late. Critically, this temporal signature of deviation from optimality predicts individual differences in self-reported procrastination independently of overall task performance. Using machine learning to relate behavioral patterns across 24 conditions of a sequential resource allocation task to validated self-report measures, we show that procrastinators exhibit a specific timing dysfunction: being suboptimal early when optimal policy prescribes work -- not merely performing worse on average. These findings reframe procrastination as bounded rationality under computational constraints rather than pure irrationality, suggesting that interventions targeting uncertainty reduction or temporal scaffolding may prove more effective than exhortations to increase motivation or willpower.