Weighting waiting: A decision-theoretic taxonomy of delay, pacing and procrastination

Sahiti Chebolu, Peter Dayan

TL;DR

This paper introduces a novel decision-theoretic framework to unify the study of pacing, delay, and procrastination. By modeling these behaviors as a sequence of temporal decisions, the authors propose a taxonomy of mechanisms that explain why and how individuals choose to postpone or rush tasks, offering a more comprehensive view than previous models focused solely on temporal discounting.

What Problem Does This Paper Address?

We all face daily decisions about when to start tasks, how much effort to exert, and how to pace our work over time. While procrastination – delaying tasks despite negative consequences – is a common and detrimental behavior, existing research often treats pacing and procrastination separately and relies heavily on temporal discounting (valuing immediate rewards more than future ones) as the primary explanation. This paper argues that this view is too narrow. It aims to bridge the gap between pacing and procrastination research by developing a unified, decision-theoretic framework that accounts for a broader range of psychological and situational factors influencing these temporal decisions.

Methods

The authors employ a Markov Decision Process (MDP) framework to formalize the decision-making processes involved in pacing and procrastination. This involves defining states (e.g., task progress, time to deadline), actions (e.g., amount of effort to apply), and rewards/costs. They then derive optimal and sub-optimal policies that dictate an individual's behavior. The core of their approach is a taxonomy of pacing types based on how individuals plan and execute tasks:

  1. Sticking to an open-loop plan: Committing to a plan and executing it without deviation.
  2. Defecting on a plan: Deviating from an intended plan, often involving delays.
  3. Making a contingent plan (closed-loop): Planning actions based on anticipated future circumstances.

They then explore various mechanisms (e.g., discounting, mis-estimation of efficacy, fatigue, non-linear effort costs) that can lead to these pacing types, simulating them using computational models to demonstrate how they can generate different pacing styles and, in some cases, procrastination.

Datasets

Not discussed in the paper. The paper focuses on theoretical modeling and simulations rather than empirical data analysis from specific datasets.

Key Insights and Ideas

  • Unified Framework: The paper successfully integrates diverse mechanisms for pacing and procrastination under a single, decision-theoretic umbrella, moving beyond the sole reliance on temporal discounting.
  • Decision Sequence Matters: The taxonomy highlights that how decisions are made (planning, defecting, adapting) is crucial for understanding pacing and procrastination, not just the outcome of those decisions.
  • Sub-optimality as a Driver: The research emphasizes that procrastination often arises not from irrationality, but from sub-optimal decision-making processes (e.g., mis-estimating abilities, temporal inconsistencies) that lead to negative consequences.
  • Beyond Discounting: The paper identifies several mechanisms beyond temporal discounting (e.g., fatigue, effort costs, environmental changes) that significantly influence pacing and delay.
  • Pacing Styles as Outcomes: Different pacing styles (early, steady, deadline, U-shaped) emerge naturally from the interplay of these mechanisms and decision-making strategies.

Key Results

The paper presents a theoretical taxonomy and demonstrates its plausibility through simulations of six distinct models. These simulations show how various mechanisms can lead to:

  • Different pacing styles: From early completion to deadline-driven work.
  • Delays in task initiation and execution: Driven by factors like discounting, fatigue, or waiting for opportune moments.
  • Procrastination: Specifically, models involving mis-estimated efficacy and steeper discounting of effort over rewards were shown to lead to sub-optimal policies resulting in negative consequences.
  • Quantitative findings: While not focused on specific accuracy metrics, the simulations illustrate how parameters like discount factors (γ), efficacy (η), and effort costs (r_effort) quantitatively influence completion times, rates, and the shape of work trajectories. For instance, lower discount factors (more impatience) lead to longer delays, and lower efficacy increases task completion time.

Limitations and Future Work

The authors acknowledge several limitations and suggest future directions:

  • Model Simplifications: The MDP framework assumes discrete states and actions, and deterministic rewards, which are simplifications of real-world complexity. Future work could explore Partially Observable Decision Processes (POMDPs) to account for uncertainty about task difficulty or skill levels.
  • Commitment Devices: The models do not explicitly include commitment devices that individuals might use to enforce plans, which could be an important extension.
  • Multi-tasking and Goal Choice: The current taxonomy primarily focuses on single-task pacing. Future research could extend this to multi-tasking scenarios and the complex decisions involved in goal selection and prioritization.
  • Experimental Validation: The authors emphasize the need for targeted experimental manipulations to test the proposed mechanisms and validate the taxonomy against empirical data. This could involve varying task characteristics (deadlines, effort, risk) and observing how pacing behavior changes.
  • Broader Applications: The framework could be extended to other forms of temporal decisions beyond task completion, and to investigate phenomena like "pre-crastination" (rushing tasks to one's detriment).
  • Online Data: The increasing availability of online activity data (e.g., from learning management systems) offers promising avenues for detailed, fine-grained analysis of pacing behavior.

Summary updated

Abstract

Why do today, what you can fail to do tomorrow? Pacing styles involving postponing, and ultimately procrastinating tasks are widespread pathologies to which many succumb. Previous research has recognised multiple types of delayed working, influenced by a myriad of psychological and situational factors. However, the main mechanistic explanation for delays, which concerns temporal discounting of distant rewards, only encompasses a subset. Further, investigations of pacing have been rather independent from those of procrastination. Here, we introduce a systematic taxonomy of pacing and procrastination within a common framework that is based on the type of temporal decisions involved in choosing to delay. We suggest that these decisions are driven by characteristics of the task and the (sub-)optimality of the decision-making. We illustrate aspects of the taxonomy by simulating diverse sources of behavioral delay using reinforcement learning models. We also analyse whether students pacing their work through a semester in a real-world task (Zhang & Ma, Scientific Reports, 2024) showany evidence of behaving according to the types detailed in our taxonomy. Our approach provides a theoretical foundation for understanding pacing and procrastination, enabling the integration of both established and novel mechanisms within a unified conceptual framework.
BibTeX
@misc{chebolu2025weighting,
  author = {Sahiti Chebolu and Peter Dayan},
  title = {{Weighting waiting: A decision-theoretic taxonomy of delay, pacing and procrastination}},
  year = {2025},
  doi = {10.31234/osf.io/jd4ar},
  howpublished = {PsyArXiv},
}

Loading related content...

Figure