Study pack · Experimental psychology and methods

Deadlines and procrastination

Experimental psychology · as it stood on 14 July 2026 · run unseen-2026-11-07/deadlines-procrastination · Claude Opus 5.5

Has a serious holeNo break

Spaced deadlines are also reminders and a pacing signal.

claims: 6
holes: 8
lenses: 3
records cited: 12 of 14
retracted precedents: 0

What happened, 2 September 2026. Psychological Science retracts it. A replication published on 15 July 2026 found that changing the deadlines had a negligible effect on performance. The journal retracted the paper on 2 September 2026. Its notice says the replication raised questions about the underlying data, and that later analyses called the overall findings into question. Asked first, the model said it didn’t know how it ended.

The idea, as it stood

A 2002 paper in Psychological Science asks whether deadlines help people overcome procrastination. In its second study, sixty participants were paid to proofread three ten-page documents over 21 days, twenty in each of three conditions: one document due each week, deadlines they chose themselves, or all three due on the last day. Participants with evenly spaced deadlines made about 136 corrections on average, against about 71 with a single final deadline, and were also less late and earned more; those who chose their own deadlines fell in between. The authors conclude that people recognise their self-control problems and impose costly deadlines on themselves, and that these deadlines help but are not as effective as evenly spaced external ones. The paper is widely taught and has been cited more than 2,000 times.

With twenty paid participants per condition, evenly spaced deadlines produced about 136 corrections on average against about 71 under a single final deadline, along with less lateness and higher earnings.empirical · supported
Evenly spaced deadlines raised performance because they overcame procrastination, a failure of self-control, rather than through some other feature of the schedule.causal · contested
Participants who set their own deadlines did so because they recognised their own self-control problems.causal · contested
The deadlines participants set for themselves were genuinely costly, so choosing them cannot be explained without a self-control motive.empirical · unsupported
Self-chosen deadlines improve performance over a single final deadline but less than evenly spaced external deadlines, and this in-between ordering is a real effect rather than noise in samples of twenty.causal · unsupported
Results from sixty paid proofreaders over 21 days generalise to how people in general procrastinate and respond to deadlines.empirical · contested

The holes, and who found them

More than one lens found a hole in claim 1, each on its own. That is a stronger signal than one lens alone.

#	Hole	Lens	How serious	Claim	Status
1	Spaced deadlines are also reminders and a pacing signal	Herbert Simon	serious	2	Open
2	Working late may have been a plan, not a lapse	Herbert Simon	serious	2	Open
3	Twenty per group: the gap's size is loose and the replication's result is missing	Richard Feynman	serious	1	Open
4	A self-set deadline you expect to meet costs almost nothing	John von Neumann	serious	4	Open
5	'In between' would appear half the time if self-set deadlines did nothing	Richard Feynman	serious	5	Open
6	The motive was inferred from the choice, never measured	Richard Feynman	serious	3	Open
7	One paid task in three equal parts is a narrow environment	Herbert Simon	serious	6	Open
8	Three agreeing outcomes are not three confirmations	John von Neumann	minor	1	Open

What it rests on

Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""Unknown authors · 2025 · Publons reviews and discussion · cited on claim 1 · doi:10.1177/09567976261460772/v2/review1
Present-Bias, Procrastination and Deadlines in a Field ExperimentBisin & Hyndman · 2014 · National Bureau of Economic Research · cited on claim 2, 3, 4, 6 · doi:10.3386/w19874
Demand and Supply of Infrequent Payments as a Commitment Device: Evidence from KenyaCasaburi & Macchiavello · 2019 · American Economic Review · cited on claim 3, 4 · doi:10.1257/aer.20180281
On the Interaction of Memory and Procrastination: Implications for RemindersEricson · 2014 · National Bureau of Economic Research · cited on claim 2, 4 · doi:10.3386/w20381
Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""Unknown authors · 2024 · Publons reviews and discussion · cited on claim 1 · doi:10.1177/09567976261460772/v1/review1
Self-Regulation of Slippery Deadlines: The Role of Procrastination in Work PerformanceSteel et al. · 2022 · Frontiers in Psychology · cited on claim 6 · doi:10.3389/fpsyg.2021.783789
Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""Unknown authors · 2024 · Publons reviews and discussion · cited on claim 1 · doi:10.1177/09567976261460772/v1/review2
Healthcare Appointments as Commitment DevicesDerksen et al. · 2024 · The Economic Journal · cited on claim 2, 5, 6 · doi:10.1093/ej/ueae077
Deadlines, Procrastination, and Forgetting in Charitable Tasks: A Field ExperimentKnowles et al. · 2017 · RePEc: Research Papers in Economics · cited on claim 2, 6 · record
Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""Unknown authors · 2025 · Publons reviews and discussion · cited on claim 1 · doi:10.1177/09567976261460772/v2/review2
Self Control and Commitment: Can Decreasing the Liquidity of a Savings Account Increase Deposits?Beshears et al. · 2015 · National Bureau of Economic Research · cited on claim 3 · doi:10.3386/w21474
Deadlines and Memory LimitationsAltmann et al. · 2021 · Management Science · cited on claim 2, 6 · doi:10.1287/mnsc.2021.4227
Procrastination, Deadlines, and Statutes of LimitationWistrich · 2008 · The William & Mary Law School Scholarship Repository (William & Mary) · retrieved, not cited · record
Do Civil Servants Respond to Behavioral Interventions?: A Field ExperimentScartascini & Zamora · 2021 · retrieved, not cited · doi:10.18235/0003753

The tests that settle it

The deciding test · months of work

A preregistered rerun with about 100 participants per arm, five times the original, and three arms: evenly spaced binding deadlines, a single final deadline, and a single final deadline with plain reminders on days 7 and 14 that the work is due on day 21, naming no intermediate dates. Every arm answers the same day-one questions: when, given their other commitments, they would choose to work on each document if they could hold themselves to it, and when they expect they will. Actual work dates are logged. Compare corrections across arms, and actual against chosen dates within each arm.

It fails if: The plain-reminder arm recovers more than half of the corrections gap between the spaced and final-deadline arms, or actual work in the final-deadline arm falls no further after participants' own chosen dates than in the spaced arm: either means overcome self-control is not what drives most of the effect. If the spaced-over-final gap does not appear at all, the empirical core fails as well.

For hole	Test	Result
1	Add an arm with a single final deadline plus plain reminders on days 7 and 14 that the work is due on day 21, naming no intermediate dates (a reminder that names dates can act as a soft commitment, as E9's appointments did), and log when work is done. If the plain-reminder arm recovers more than half of the corrections gap between the spaced and final-deadline arms, memory and attention rather than overcome self-control carry most of the effect; if it recovers little while binding spaced deadlines recover all of it, the self-control reading gains.	Proposed
2	On day one, ask every participant in every arm when, given everything else due in those weeks, they would choose to work on each document if they could hold themselves to it, and when they expect they actually will; log when each document is worked on. Self-control failure predicts that in the final-deadline arm actual work falls well after those chosen dates, foreseen or not, and that the shortfall in corrections tracks the delay. It counts against the idea if actual timing in the final-deadline arm matches participants' own chosen dates as closely as in the spaced arm.	Proposed
3	From the replication's data or reported statistics, compute its spaced-minus-final difference in standard-deviation units with a 95% interval, and the original's from its reported spread. It counts against the claim if the replication's interval includes zero or excludes the original's estimate.	Proposed
4	First report how many self-set deadlines were missed and what penalties were actually paid: that is the realised cost. Then, in a new arm, attach a small certain fee to each early binding deadline, or offer a free non-binding planning calendar beside it. A commitment motive predicts people keep paying for binding early deadlines, as E3's farmers paid for commitment; it counts against the claim if demand for early binding deadlines largely vanishes once it carries a small certain price or sits beside a free plan.	Proposed
5	From the per-participant data, compute both contrasts the claim needs, self-chosen minus final and spaced minus self-chosen, each with a 95% interval; the in-between claim stands only if both exclude zero. To separate weaker deadlines from patchy uptake, add an arm in which everyone must choose three separate deadlines of their own spread across the 21 days: it counts against 'less effective' if that arm matches the spaced arm.	Proposed
6	Before the deadline choice, ask each participant when they would choose to work on each document if they could hold themselves to it and when they expect they actually will, and measure present bias, for example with incentivised choices between sooner and later payments. Recognising a self-control problem means expecting to work later than one would choose. It counts against the claim if neither that expected gap nor measured present bias predicts who sets early deadlines.	Proposed
7	Repeat the three arms with a different population, a task with unequal parts and no performance pay, and a longer window. It counts against generalisation if the spaced-over-final advantage shrinks to nothing or reverses there.	Proposed
8	Recompute each participant's pay from the payment rule using their corrections and lateness. If that reproduces the earnings gap, earnings should be reported as derived, and it counts against presenting the result as three separate findings.	Proposed

Proposed tests that haven’t run are the agenda for what comes next. Download them as a draft pre-registration, ready to edit and register on OSF.

Two answers, matched blind

The same model answered twice: once plainly, as a standard AI, and once through Feynman. Both answers’ findings were matched blind, without the ending.

The lenses found, the standard AI missed

Spaced deadlines are also reminders and a pacing signal Simon's idea that attention is the scarce resource led the lens to read each deadline as a reminder and pacing cue, an account needing no self-control failure. Its cited post-1980 studies added evidence on memory and forgetting, and one commitment result. The standard AI instead proposed fatigue or extra effort, and doubted early choices show self-knowledge.

The standard AI found, the lenses missed

The middle result could be a mix of people, not a weaker tool The standard AI cited nothing; it reasoned that a middle average could hide a split, some choosing spaced dates and doing well, others late dates and doing badly, which within-condition comparison cannot settle. Feynman instead held that 'in between' would appear half the time if self-set deadlines did nothing, and that the motive was never measured.

Fame is not replication, and the data deserve a check The standard AI set the paper's fame against knowing of no large preregistered repeat, and cited a later paper co-authored by the first author, retracted in 2021 over fabricated data, as a reason to check raw data, not evidence against this one. Feynman raised replication only as a missing result, keeping to the study's design.

The finding

Spaced deadlines plausibly raised output, but the design cannot separate self-control from reminders, pacing and planned lateness, the self-set deadlines may have cost almost nothing to choose, and the in-between result rests on groups of twenty.

It failed: Psychological Science retracts it. What would overturn the review’s reading: The plain-reminder arm recovers more than half of the corrections gap between the spaced and final-deadline arms, or actual work in the final-deadline arm falls no further after participants' own chosen dates than in the spaced arm: either means overcome self-control is not what drives most of the effect. If the spaced-over-final gap does not appear at all, the empirical core fails as well.

Cite it, disclose it, reproduce it

A paragraph for your methods section

We used Feynman (feynman.works), an AI research-review tool running on Claude Opus 5.5, to critique “Deadlines and procrastination” on 2026-11-07 (run unseen-2026-11-07/deadlines-procrastination). It reconstructed 6 claims, retrieved 14 published records dated no later than 2026-07-14, and raised 8 objections through reasoning frameworks drawn from the published work of Richard Feynman, Herbert Simon and John von Neumann. Each citation in its output was checked against the records it retrieved, and we read every source we relied on. Its objections informed our analysis; the conclusions, and any errors, are our own.

The lenses

Richard FeynmanPhysics & scientific method · 1948–1980What experiment would distinguish this explanation from a convincing story?
Herbert SimonDecision science · 1947–1980How does a person decide with limited time, attention and information?
John von NeumannMathematics & strategy · 1928–1957What are the incentives, available strategies and consequences of each move?

BibTeX

@misc{feynman-unseen-2026-11-07-deadlines-procrastination,
  title = {{Feynman review: Deadlines and procrastination}},
  author = {{Feynman Works}},
  year = {2026},
  howpublished = {\url{https://www.feynman.works/pack/deadlines-procrastination}},
  note = {Run unseen-2026-11-07/deadlines-procrastination, Claude Opus 5.5, 2026-11-07}
}

@article{unknown2025review,
  title = {{Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""}},
  author = {Unknown authors},
  year = {2025},
  journal = {Publons reviews and discussion},
  doi = {10.1177/09567976261460772/v2/review1}
}

@article{bisin2014present,
  title = {{Present-Bias, Procrastination and Deadlines in a Field Experiment}},
  author = {Bisin and Hyndman},
  year = {2014},
  journal = {National Bureau of Economic Research},
  doi = {10.3386/w19874}
}

@article{casaburi2019demand,
  title = {{Demand and Supply of Infrequent Payments as a Commitment Device: Evidence from Kenya}},
  author = {Casaburi and Macchiavello},
  year = {2019},
  journal = {American Economic Review},
  doi = {10.1257/aer.20180281}
}

@article{ericson2014the,
  title = {{On the Interaction of Memory and Procrastination: Implications for Reminders}},
  author = {Ericson},
  year = {2014},
  journal = {National Bureau of Economic Research},
  doi = {10.3386/w20381}
}

@article{unknown2024review,
  title = {{Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""}},
  author = {Unknown authors},
  year = {2024},
  journal = {Publons reviews and discussion},
  doi = {10.1177/09567976261460772/v1/review1}
}

@article{steel2022self,
  title = {{Self-Regulation of Slippery Deadlines: The Role of Procrastination in Work Performance}},
  author = {Steel and others},
  year = {2022},
  journal = {Frontiers in Psychology},
  doi = {10.3389/fpsyg.2021.783789}
}

@article{unknown2024reviewb,
  title = {{Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""}},
  author = {Unknown authors},
  year = {2024},
  journal = {Publons reviews and discussion},
  doi = {10.1177/09567976261460772/v1/review2}
}

@article{derksen2024healthcare,
  title = {{Healthcare Appointments as Commitment Devices}},
  author = {Derksen and others},
  year = {2024},
  journal = {The Economic Journal},
  doi = {10.1093/ej/ueae077}
}

@article{knowles2017deadlines,
  title = {{Deadlines, Procrastination, and Forgetting in Charitable Tasks: A Field Experiment}},
  author = {Knowles and others},
  year = {2017},
  journal = {RePEc: Research Papers in Economics},
  url = {https://mpra.ub.uni-muenchen.de/83694/1/MPRA_paper_83694.pdf}
}

@article{unknown2025reviewb,
  title = {{Review for "Replication of “Procrastination, Deadlines, And Performance: Self-Control by Precommitment""}},
  author = {Unknown authors},
  year = {2025},
  journal = {Publons reviews and discussion},
  doi = {10.1177/09567976261460772/v2/review2}
}

@article{beshears2015self,
  title = {{Self Control and Commitment: Can Decreasing the Liquidity of a Savings Account Increase Deposits?}},
  author = {Beshears and others},
  year = {2015},
  journal = {National Bureau of Economic Research},
  doi = {10.3386/w21474}
}

@article{altmann2021deadlines,
  title = {{Deadlines and Memory Limitations}},
  author = {Altmann and others},
  year = {2021},
  journal = {Management Science},
  doi = {10.1287/mnsc.2021.4227}
}

Run unseen-2026-11-07/deadlines-procrastination on Claude Opus 5.5. The review is reproducible from its record: the claims, the records retrieved before 14 July 2026, and each lens’s card. https://www.feynman.works/pack/deadlines-procrastination