Decision Fatigue

Also called Choice fatigue

Quick answer

Decision fatigue is the claim that choices get worse after a person has made many decisions in a row. Tired decision makers more often pick the easiest option: the default, a refusal, or a delay. It is usually explained by a limited supply of self-control that each choice uses up, an explanation that researchers still dispute.

~20 sec read

011 min

Where decision fatigue shows up

Say a product manager spends a Thursday triaging bug reports. For each one they decide whether it ships this sprint, waits, or is closed. By late afternoon they have made about forty of these calls.

The forty-first report describes a rare path that loses customer data. Judging it well means reading server logs and talking to an engineer. The product manager types "revisit next week" into the tracker and opens the next report.

Now change one input. The same report arrives at nine the next morning, before any other triage. The product manager reads the logs, messages the engineer, and puts the fix into the sprint.

The report is identical in both versions. Only the number of earlier decisions differs. The option that needed the most effort, investigating, is the one that was dropped. Postponing costs nothing today, so it is chosen more often late in a long run of decisions.

That is the pattern the label points to. After many choices in a row, people pick the option that asks the least of them. The least demanding option is usually the default, a refusal, or a delay.

021 min

Where decision fatigue comes from

The most cited test of the idea is Vohs, Baumeister and colleagues (2008), published in the Journal of Personality and Social Psychology. It reported four laboratory studies. In each, some participants made choices among consumer goods or college course options. Others only thought about the same options without choosing.

Afterwards, everyone did tasks that needed self-control. The people who had chosen showed less physical stamina, gave up sooner after failing, put things off more, and did arithmetic with less quality and quantity.

The authors added a field study. Shoppers who said they had made more active decisions earlier showed less self-control later.

Two further results shaped the idea. Choosing was more depleting than only deliberating and forming preferences. It was also more depleting than carrying out a choice that someone else had made.

The paper's reasoning rests on ego depletion, the proposal that self-control draws on one limited supply and that effort of any kind lowers it. Making choices, the authors argued, uses the same supply. Later medical researchers describe "the erosion of self-control after making repeated decisions" as decision fatigue, and they cite this paper.

Replication is part of the story. The ego-depletion proposal behind this work has not held up well in large tests, which the section on contested evidence covers.

032 min

Why decision fatigue is said to happen

Two explanations compete. Both predict the same behaviour. They differ on what restores the ability to choose well.

The resource account. In this account, self-control is a limited supply. Resisting a temptation, holding back an emotion, and making a choice each use some of it. When the supply is low, effortful options become harder to take. Comparing options carefully, resisting an impulse, and acting on one's own initiative all cost effort. The cheap options, such as the default or doing nothing, cost almost none. Under this view, rest or food restores the supply.

The motivation account. Inzlicht, Schmeichel and Macrae (2014) argue that nothing is used up. They propose that apparent failures of self-control come from "the motivated switching of task priorities". After sustained effort, people shift from "have-to" goals toward "want-to" goals and prefer what they call cognitive leisure. Under this view, a new goal or a reward can restore effort even with no rest.

The motivation account also gives a reason the behaviour exists. Mental effort has a price, and people weigh it against other things they could be doing. An option that saves effort becomes more attractive as the effort already spent grows.

Both accounts agree on one practical prediction. Any long run of effortful, similar decisions with no break and no change of task is a place where the cheapest option will be chosen more often. A reviewer scoring two hundred job applications in one sitting is one such place. That is a prediction from the accounts, not a measured result for that setting.

Neither account is settled. The resource account depends on ego depletion, which a large replication did not confirm. The motivation account is newer and has been tested less.

041 min

Examples

A real case: parole boards in Israel

Danziger, Levav and Avnaim-Pesso (2011) studied sequential parole decisions by experienced judges in Israel. They recorded each judge's two daily food breaks. The breaks split the working day into three decision sessions.

The result was a repeating pattern. The share of favourable rulings fell gradually from about 65% to nearly zero within each session. It returned abruptly to about 65% after a break.

The authors concluded that judicial rulings "can be swayed by extraneous variables that should have no bearing on legal decisions". The cost, if the pattern is real, is borne by prisoners whose cases came late in a session. A fix would be to schedule breaks so that no case is heard at the end of a long run.

The next section on contested evidence explains why this reading is disputed.

A design critique

A design lead runs a review of twenty-five concept boards in one afternoon. For each board they choose between "approve", "revise" and "reject". Late in the session they approve boards faster and ask fewer questions. Approving ends the discussion at once.

The decision point here is the order of the boards. Putting the riskiest board first, or splitting the review over two sessions, changes which boards get the careful look. The definition of decision fatigue says nothing about ordering as a design choice. The scenario adds it.

052 min

A second case: antibiotic prescriptions

Linder and colleagues (2014) studied a different setting. They merged billing and health-record data from 23 primary care practices in one health system. They looked at visits by adults with acute respiratory infections. They kept visits in two four-hour sessions each day, 8 AM to noon and 1 PM to 5 PM.

The study covered 21,867 visits to 204 clinicians, and 44% of visits ended in an antibiotic prescription. Prescribing rose through each session. Compared with the first hour, the odds of a prescription were 14% higher in the third hour and 26% higher in the fourth hour. This held for infections where antibiotics are sometimes indicated and for those where they are never indicated.

The authors list reasons clinicians might prescribe when it is not needed: patient demand, a wish to do something, a wish to finish the visit quickly, and fear of complications. Prescribing is the easy option in that setting.

The two cases differ in one variable. In the parole study, the more frequent late outcome was a refusal. In the clinic study, it was an action, a prescription. So decision fatigue does not mean "say no" or "say yes". It means the option that is easiest in that setting becomes more likely.

Linder's authors also name limits. Scheduled time stood in for real visit time, and other things that build up during a session, such as low blood glucose or general tiredness, could account for the result.

062 min

Where the evidence is contested

The parole result has a named critic. Weinshall-Margel and Shapard (2011) examined the original data, collected more from 12 hearing days, and interviewed lawyers, a parole panel judge and court staff. They found that "case ordering is not random".

Three details of the ordering matter. The board tries to finish all cases from one prison before it takes a break. Within each session, prisoners without a lawyer usually go last, and they are less likely to be granted parole. The authors also suspect that lawyers present their best cases first and keep their weakest for last.

Their strongest objection is that this ordering alone could produce a falling success rate inside each session, with no tiredness at all. Danziger and colleagues published a reply defending their result. This article has not weighed the reply's details, so treat the parole case as open.

The wider theory is contested too. Hagger and colleagues (2016) ran a preregistered replication of ego depletion in 23 laboratories with 2,141 participants in total. The effect was small, and its 95% confidence interval included zero (d = 0.04). The study used a different pair of tasks from choosing, so it does not decide whether choosing in particular uses up anything. It does weaken the resource account that decision fatigue was built on.

What should a practitioner do while this is unsettled? Treat the pattern as a plausible risk, not a law. Cheap changes such as scheduling the hardest decisions first and adding breaks cost little if the effect is smaller than reported. Expensive changes, such as removing human judgment, are not justified by this evidence.

072 min

How decision fatigue shows up in product, design and AI

No study cited here tested the three settings below. They are places with the same structure as the parole and clinic studies: a long run of similar decisions, made by one person, where one option takes least effort.

  • A product backlog review. A product manager reviews eighty tickets in one meeting. "Not now" is the easiest label for each. Late tickets get it more often, whatever their value.
  • Code review. An engineer reviews a queue of pull requests in one sitting. Approving with a short comment takes less effort than tracing a change through the system. The effect is accidental, not designed.
  • Rating model outputs. An analyst scores two hundred answers from a language model against a rubric. The easy pattern is to give the middle score. Late in the batch, middle scores may become more common. This is a risk to check in the data, for example by comparing score distributions early and late in a session.

In all three, the effect would be accidental. Nobody designs a review queue to make its reviewers choose the cheapest option.

The designer's version is about users. A form or a setup flow that asks many questions in a row may lead people to accept whatever is pre-selected near the end. Whether it does is a testable question. A team can compare how often people keep the pre-selected answer on early screens and on late screens.

081 min

How to guard against decision fatigue

Four changes follow from the pattern. The first, third and fourth are reasoned from the studies, not tested remedies.

  1. Put the hardest decisions first.

    If care declines over a session, give your most careful attention to the choices that matter most. In the Thursday example, the data-loss report goes first.

  2. Break the run.

    The parole study found that favourable rulings returned to their starting level after a break. Linder and colleagues suggest "time-dependent decision support, modified schedules, shorter sessions, mandatory breaks, or snacks". These are proposals, and the authors did not test them.

  3. Decide the rule in advance.

    A written rubric or a checklist moves the effort from the moment of choice to the moment of writing the rule. A rule that fixes a default in the safe direction also protects the tired decider.

  4. Make the easy option take a step.

    If "postpone" or "reject" is the default, require one line of reasoning to choose it. Choice architecture is the field that studies how the layout of options changes which one gets picked.

None of the studies cited here tested whether knowing about decision fatigue changes behaviour. Plan around the structure of the day, not around good intentions.

091 min

Common misunderstandings

"Every decision uses up willpower." Vohs and colleagues found that choosing was more depleting than deliberating and forming preferences, and more depleting than carrying out a choice made by someone else. So the claim was never that thinking about options tires people equally. It was about the act of choosing.

"Willpower is a muscle that tires and recovers." This picture comes from the resource account. The registered replication in 23 laboratories did not support it for the task it used, so it should not be stated as fact.

"The parole study proves hungry judges are harsh." The study measured the position of a case in a session, not hunger or blood glucose. The critique above gives a different explanation for the same numbers.

"Fewer decisions is always better." Removing choices helps when the choices are routine, such as a standing order for a repeated task. It does harm when cases differ and judgment is the point. A parole board that decided by a fixed rule would end the late-session pattern and would also stop weighing individual cases.

101 min

Decision fatigue vs. nearby concepts

Four neighbours are often mixed up with decision fatigue. Each differs on one axis.

ConceptWhat it is aboutThe axis that separates it
PsychologyChoice overloadNumber of options at once, not number of decisions over timeWhat it is about: Too many options in one choice
PsychologyAnalysis paralysisThe person stalls, whereas with decision fatigue they decide quickly and cheaplyWhat it is about: Failing to decide at all
Not in the library yetCognitive loadDemand in the moment, not the accumulated effect of earlier decisionsWhat it is about: The mental demand of one task right now
PsychologyAkrasiaA general failure of will, which decision fatigue may cause but does not defineWhat it is about: Acting against one's own best judgment

The simplest test is timing. If the problem appears because of one large menu, it is choice overload. If it appears because of what came before the menu, it is decision fatigue.

?6 questions

Questions people ask

What causes decision fatigue?

No single cause is established. The older account says each choice uses part of a limited supply of self-control. A newer account says that after effort, people switch from have-to goals to want-to goals. Both predict that the easiest option is chosen more often late in a run of decisions.

What is an example of decision fatigue at work?

A named example is the 2011 study of Israeli parole boards, where favourable rulings fell from about 65% to nearly zero within each session. That reading is disputed. A second example is clinic data in which antibiotic prescribing for respiratory infections rose through each four-hour session.

How do you avoid decision fatigue?

Put the hardest decisions at the start of a session, take breaks, and decide rules in advance so fewer choices need effort. Linder and colleagues propose mandatory breaks and shorter sessions. These are reasoned proposals that no study cited here has tested.

What is the difference between decision fatigue and choice overload?

Choice overload comes from too many options in a single choice. Decision fatigue comes from the number of decisions that came before. A person can have either one without the other.

Is decision fatigue real?

The pattern appears in some data, but its cause is unsettled. A 23-laboratory preregistered replication of the ego-depletion effect found an effect near zero, and a critique offers a different explanation for the parole result. Treat it as a plausible risk, not a proven law.

Does eating fix decision fatigue?

Not reliably. Linder and colleagues list snacks among possible remedies and name low blood glucose as one possible factor, but they did not test either. The parole study recorded food breaks and did not measure glucose.

§6 sources

Sources

Show all 6 sources
  1. Danziger, S., Levav, J. and Avnaim-Pesso, L. (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences 108(17), 6889-6892

  2. Weinshall-Margel, K. and Shapard, J. (2011). Overlooked factors in the analysis of parole decisions. Proceedings of the National Academy of Sciences 108(42), E833

  3. Vohs, K. D., Baumeister, R. F., Schmeichel, B. J., Twenge, J. M., Nelson, N. M. and Tice, D. M. (2008). Making choices impairs subsequent self-control. Journal of Personality and Social Psychology 94(5), 883-898

  4. Linder, J. A., Doctor, J. N., Friedberg, M. W., Reyes Nieva, H., Birks, C., Meeker, D. and Fox, C. R. (2014). Time of day and the decision to prescribe antibiotics. JAMA Internal Medicine 174(12), 2029-2031

  5. Hagger, M. S., Chatzisarantis, N. L. D. et al. (2016). A multilab preregistered replication of the ego-depletion effect. Perspectives on Psychological Science 11(4), 546-573

  6. Inzlicht, M., Schmeichel, B. J. and Macrae, C. N. (2014). Why self-control seems (but may not be) limited. Trends in Cognitive Sciences 18(3), 127-133

Keep reading

More from Psychology

All of Psychology
All of Psychology

Missing a concept?

Tell us what to write next. We’ll email you when it’s live.

Several? Separate them with commas.

Only to tell you when it’s live.

How we use your email: privacy.