Decision Fatigue: How to Decide Well When You Are Exhausted

It is nine in the evening, the same question is in front of you for the third time today, and you can feel yourself reaching for whichever answer ends the conversation fastest. You already know the phrase for this. Everyone does.

The phrase is decision fatigue, and the popular version of it is oversold. The research underneath the advice is considerably shakier than the advice admits, which would be a footnote if the story did not also tell you what to do about it. Fix the wrong thing and you will spend real effort and still make the same poor call at nine in the evening.

The experience is real. Late in a hard day, people decide worse. What is probably wrong is the explanation, and the explanation is what determines the remedy.

What the Famous Study Actually Found

Almost every article on this subject leans on one piece of evidence: a 2011 paper in PNAS by Shai Danziger, Jonathan Levav and Liora Avnaim-Pesso, looking at rulings by Israeli parole boards. The headline number is the one you have seen. Favourable rulings ran at roughly 65 per cent at the start of a decision session and fell to near zero just before the board broke for food, then jumped back up afterwards.

It is a wonderful finding. It is also one of the most contested results in behavioural science, and that part almost never travels with it.

Kamila Weinshall-Margel and John Shapard responded in the same journal with an awkward observation about how those hearings are scheduled. The order of cases is not random. Prisoners without legal representation tend to be heard last in a session, and prisoners sharing an attorney can be grouped by that attorney’s convenience. If the cases most likely to be refused cluster at the end of a session for reasons of scheduling, you get the graph without any tired judges in it at all.

Andreas Glöckner went further in Judgment and Decision Making in 2016, with a simulation of a completely unbiased judge. Favourable decisions take longer to process. A judge who will not start a case they cannot finish before the break will tend to slot the quick refusals into the last few minutes. That alone reproduces a decline that looks like the real thing. Glöckner also pointed out the size problem: the parole effect works out at around d = 1.96, more than twice the conventional threshold for a large effect in psychology, while meta-analytic estimates of mental depletion effects sit somewhere between about d = -0.10 and d = 0.25. An effect roughly ten times larger than the mechanism that supposedly causes it is not impossible, but it should make you want a second look.

The original authors replied and stood by their reading. Nobody has proven them wrong. What has happened is that a finding presented to the public as settled turns out to be a live argument among the people who study it, and you were never told.

The Mechanism Everybody Assumed

Behind decision fatigue sits a bigger idea called ego depletion: the theory that self-control draws on a limited shared resource, so that resisting a biscuit at eleven leaves you with less in the tank for a hiring decision at four. It is a wonderfully intuitive picture. It feels exactly like what being tired is like, which is part of why it spread so fast.

In 2016, twenty-three laboratories ran the same preregistered protocol on 2,141 participants to test it properly. The multilab registered replication published in Perspectives on Psychological Science found an effect of d = 0.04, with a confidence interval comfortably containing zero. In plain terms, they looked hard for the battery and did not find it.

This is contested too. Roy Baumeister, whose work the theory came from, objected that the replication used a weak version of the depleting task. That objection is worth taking seriously, and the honest summary is that the effect, if it exists, is far smaller and far more fragile than a decade of popular writing suggested. The related story about willpower running on blood glucose, the one that ends with advice to eat something sugary before a hard conversation, rests on the same shaky footing and should be treated as an open question rather than a fact.

So the mechanism is in doubt. The tiredness survives that doubt untouched, which leaves a better question than the one most people ask.

What Tiredness Actually Does to a Decision

Think about what deciding actually costs you. The cheap part is picking between two options once they are sitting side by side with their trade-offs visible. The expensive part is everything before that: noticing there is a choice at all, generating the options nobody has mentioned, working out what you actually care about here, and holding three incompatible futures in your head long enough to compare them.

That is the part that goes first when you are exhausted. Your integrity holds. Your intelligence holds. What thins out is the willingness to do expensive cognitive work when a cheap path is sitting right there.

And there is always a cheap path. Four of them, usually. You take the default that is already sitting there. You keep the status quo because changing it requires justifying the change. You go with the last thing that was suggested, because evaluating it against alternatives means generating alternatives. Or you refuse everything, because no is the only answer that requires no analysis.

Each of those is a perfectly sensible energy-saving move. Each is also a decision, made by default rather than on purpose, and that is where the damage comes from.

Find the Failure Mode You Have

Here is why “make fewer decisions” and “wear the same shirt every day” miss. They treat decisions as a quantity to be rationed. Your shirt was never the problem. Nobody ever ran out of judgement because of a shirt.

The useful move is to work out which of those cheap paths you personally fall down, because they fail in completely different directions and need opposite fixes.

If you default to the status quo when tired, then every decision you take late in the day is a quiet vote to change nothing. That has a clean structural fix. Anything requiring an active change of course goes in the morning. Budget reallocations, personnel moves, the decision to stop a project that is limping along. Put those before lunch, permanently, and treat a late-afternoon slot for them as a scheduling error rather than a test of character.

If you default to yes because yes ends the conversation, the fix is a standing delay on anything agreed after a certain hour. Commitments made after six get confirmed in writing the next morning, and everyone who works with you knows it. That is not a rule against deciding late. It is a rule about when the answer counts, and it removes the incentive to keep pushing you until you fold.

If you default to no, you are probably protecting yourself from evaluation cost and calling it discipline. The tell is a run of refusals you cannot explain the reasoning behind a week later.

Decide the Rule When Fresh and the Instance When Tired

This is the single most useful habit in the whole subject, and it follows directly from what is actually expensive.

Choosing between two named options is cheap. Working out the criteria those options should be judged against is expensive. So separate them in time. When you are fresh, decide what would make a candidate a yes, what discount is worth a longer contract, what a project has to show by week six to keep going. Write the criterion down while your thinking is good. Then, when the actual case lands on a Thursday evening, you are performing a comparison rather than building a value system from scratch at the worst possible hour.

Most of what feels like decision fatigue is re-deriving the same principles over and over because nobody wrote them down the first time. What exhausts people is deciding the same matter repeatedly and getting a slightly different answer each time.

Written procedures fail for the same reason. The ones that get followed, as we found when we looked at standard operating procedures that people actually use, are the ones that record the rule and the judgement call, so nobody has to re-derive either at nine in the evening.

Write Down Why, While You Still Remember

Everyone records conclusions. Almost nobody records reasoning, and the reasoning is the part that evaporates.

A conclusion without its reasoning is worse than useless when you come back to it, because it invites the whole argument again. You look at the choice you made three weeks ago, cannot reconstruct what you knew or what you weighed, and so you re-open it. That is where the day actually goes, into re-decisions rather than decisions.

Two lines is enough. What you chose, the two or three things that moved you, and what would have to change for you to revisit it. That last clause is the valuable one, because it converts a permanent open question into a closed one with a defined trigger. It also protects you from the tired version of yourself who reopens settled matters at eleven at night for no reason beyond anxiety.

Reversible and Irreversible Are Different Problems

A great deal of exhaustion comes from treating every choice as though it were consequential. It is a strange kind of self-flattery, and it is expensive.

Most decisions are cheap to undo. Wrong vendor, switch in a quarter. Wrong meeting structure, change it next month. For these, the cost of deliberating exceeds the cost of being wrong, and the correct move when tired is to decide quickly and move on. Speed is the right answer precisely because the stakes are low.

A recurring fee is a good test case. An annual card fee looks like a big decision and is one of the cheapest to reverse, since you cancel it and the money stops. When we worked through whether the Amex Platinum is worth its $895 fee, the answer mattered less than noticing which bucket the question belonged in.

Then there is the other category. The ones with a long tail, a signed commitment, a public announcement, or a sum of money you cannot get back. Exhaustion is a genuine reason to slow down on those and only those, because that is where defaulting to the path of least effort costs you something you cannot recover. This is the same discipline behind the questions worth asking before committing to something expensive and hard to reverse, whether it is a qualification, a hire or a market you will be stuck in for three years.

Sorting choices into those two buckets takes about four seconds and saves more energy than any amount of wardrobe minimalism. Most people burn their best thinking on the reversible ones and then arrive at the irreversible one with nothing left.

Sometimes the Honest Answer Is Not Today

There is a version of this problem that no technique solves, and pretending otherwise would be dishonest.

Occasionally you are genuinely too tired to do the expensive part, the decision genuinely requires it, and no amount of pre-written criteria will get you there. The instinct in that moment is to produce a decision anyway, because producing one feels like evidence that you are still functioning and not producing one feels like a small public failure.

Say it out loud instead. I am not deciding this today, I will decide it at ten tomorrow, here is what I need in front of me when I do. That sentence costs a little social credit in the moment and saves an enormous amount later. Nearly everyone would rather wait fourteen hours for a decision that holds than get one tonight that quietly reverses itself next week.

None of this requires the willpower battery to be real. It survives the replication failures entirely, because the underlying claim is a mundane one. Thinking well is work, work gets harder when you are tired, and the sensible response to a predictable difficulty is to arrange things so the hard parts land when you can do them.

You are allowed to have believed the popular version. It was told confidently, by serious people, in places that had earned trust. The graph was compelling and the mechanism felt true from the inside. What the evidence has actually done is better news than the story it replaced. A draining battery is something that happens to you and cannot be argued with. A tendency to take the cheapest cognitive path when tired is something you can see coming, name, and build around. The first is a condition. The second is just a design problem, and you have solved harder ones before lunch.