Every choice makes the next one worse.
After a long run of decisions, people choose less carefully: they take the default, put things off, or grab whatever is easiest. Design the sequence so the decisions that matter come while attention is still fresh.
Journalist John Tierney popularized the term in 2011, drawing on research by Roy Baumeister, Kathleen Vohs and colleagues. In a 2008 series of studies, people who had just made many choices persisted less on later tasks and showed weaker self-control than people who had not.
The idea is tied to ego depletion, the theory that self-control draws on a limited resource, and that theory has struggled to replicate: a 23-lab effort in 2016 found an effect close to zero, and research with Carol Dweck suggested depletion appears mainly in people who believe willpower runs out. Field studies still show choices drifting late in long sessions.
Decision fatigue is the decline in the quality of decisions after a long session of making them. People who are worn out from choosing avoid trade-offs, put decisions off, take the default or the status quo, and become more impulsive.
In 2008 studies, people who had just made a long series of choices persisted less on later tasks and showed weaker self-control. A long run of small decisions erodes the quality of the ones that follow.
In field experiments with car buyers configuring a new car, those who started with the options that had the most choices accepted more of the defaults later on. Order decisions so the ones that matter most come first.
People who are worn out from deciding fall back on the default or the status quo. Late in a flow, the preselected option carries more weight, so make sure it is the one that serves the person.
Decision fatigue makes people plan less, procrastinate, and avoid choices altogether. Let people save progress and come back, instead of forcing every decision in one sitting.
A 2014 study found that clinicians grew more likely to prescribe antibiotics for respiratory infections as their clinic sessions went on. Decision support matters most late in a shift, when judgment is most worn down.
Candy sits by the register because shoppers reach it worn out from a store full of trade-offs. Placing upsells and add-ons where people are most depleted works, but it trades their trust for a sale.
A 2011 study reported that favorable parole rulings fell from about 65 percent to nearly zero over each session and recovered after a break. Critics pointed out that cases were not heard in random order, so treat dramatic single studies with care.
A 23-lab replication in 2016 found almost no ego-depletion effect, and other research found depletion mainly in people who believe willpower is limited. Reduce decision load because it is good design, not because a fuel-tank model of the mind is proven.
AI can take routine decisions off people’s plates, or bury them in approvals. Which one happens depends on what the system asks people to decide.
Let AI handle routine choices, such as sorting, drafting, and filling in sensible defaults, so people save their attention for the decisions that matter. Keep those decisions visible and easy to change.
AI agents that ask permission for every step train people to approve without reading. Batch routine actions, and save confirmation for the steps that are consequential or hard to undo.