Current theories of procrastination argue that people put things off into the future with the expectation that they will be better able to do them later. In this paper, we rationalize such expectations within the framework of evidence accumulation models of the choice process. Specifically, we show that it is rational for observers to adopt lower decision thresholds for choices with weak evidence for any alternative, and that observers learning to estimate optimal decision thresholds for tasks that involve decisions will find it reasonable to put the tasks off until the threshold has been sufficiently lowered by time-varying urgency. We designed a computational model and an experiment to differentiate our theory from more general expectancy based temporal motivation accounts. Both simulation and experimental results support our proposal, indicating a large role for choice difficulty in people's self-assessed estimates for how likely they are to procrastinate any given task.

It is currently difficult to test the validity of existing explana-tions for the emergence of context-dependent preference rever-sals. This is because these explanations are generally placed atthe level of the process of evidence accumulation, and acrossexperimental paradigms, this process is unobservable. In thispaper, we propose a new experimental paradigm for elicitingpreference reversals, wherein the process of evidence accumu-lation is significantly observable. Over a series of experiments,we successfully induce preference reversals for arbitrary stim-uli by showing participants sequences of stimuli comparisonswith pre-determined outcomes. Our findings partially supportthe view that context-sensitive assimilation of a history of ordi-nal comparisons is sufficient to explain classic context effects.