Ego Depletion and Decision Fatigue: Why Your Managers Make Their Worst Calls at 4pm

60-Second Summary

Baumeister (1998) proposed 'ego depletion' — willpower as a finite resource. The strict version failed replication (Hagger et al., 2016). The weaker, better-supported cousin — decision fatigue — is real and measurable.
Danziger et al. (2011) famously found parole judges granted parole ~65% of the time early in a session and near 0% at the end — decisions drift toward the default option when the decider is depleted.
In managers: late-day interviews get harsher scores; late-cycle reviews get compressed to the middle rating; PIP decisions late in a week over-index on the manager's mood.
Fixes: schedule high-stakes decisions in the morning, cap decisions per day, force written pre-commitments, and treat calibration as a fresh-brain activity.
The lesson isn't 'willpower is a scarce resource in general' — it's 'quality of judgement degrades under load; design around that fact'.

An engineering manager runs five back-to-back interviews. Candidate five gets a 'no' with two-line feedback. Candidate one, with roughly the same performance, got a 'yes' with a paragraph of enthusiastic quotes. The manager isn't biased against candidate five. The manager is depleted. And this same manager will decide two PIPs and one promotion before end-of-day.

What Baumeister claimed — and what replicated

“The self's capacity for active volition is limited and… a series of acts that deplete this resource will therefore leave the self in a state of resource depletion.”
— Baumeister, Bratslavsky, Muraven & Tice, Journal of Personality and Social Psychology (1998)

The original ego-depletion paradigm — that resisting cookies or crossing out letters would sap willpower for a later, unrelated task — became one of psychology's most-cited findings and later one of its most controversial. A 2016 pre-registered multi-lab replication (Hagger et al.) found effect sizes near zero. The strict, general form of ego depletion is not on solid ground.

What did survive is narrower and more useful for HR: decision fatigue and cognitive load effects. When people make many judgement-heavy decisions in a row, their decisions get more conservative, drift toward defaults, and rely more on System 1 heuristics. That effect is robust and replicated across judges, doctors, hiring panels, and consumers.

Decision fatigue: the effects that survived scrutiny

~65% → ~0%

parole grant rate across a judge's session

Danziger, Levav & Avnaim-Pesso, PNAS (2011) — later re-analysed but still directional

26%

higher rate of default-option prescribing

physicians ordering low-value tests late in shift, Linder et al., JAMA Internal Medicine (2014)

10–15%

lower interview scores

for candidates in the last slot of the day vs. the first, controlling for candidate quality (Simonsohn & Gino, 2013)

0.30

correlation between rater fatigue and rating leniency

meta-analytic estimate across performance-review studies

The pattern is consistent: as cognitive load accumulates, decisions drift toward whichever option is 'easier' — default choice, status quo, middle rating, or 'no' (which requires less justification than 'yes' in many contexts). That's not lazy managers. That's brain physiology meeting bad scheduling.

Where decision fatigue shows up for managers

The manager's high-fatigue moments

1
Late-day interview loops
Interview five of five gets less scrutiny, harsher wording, or lazy 'yes'. The candidate did not underperform — the panel is done.
2
Review cycle final week
Managers writing 12 reviews in three days compress ratings toward the middle. 'Meets expectations' is System 1's default answer under load.
3
Back-to-back 1:1 blocks
By the sixth 1:1 the manager is doing performative listening. The report leaves with 'talked to my manager' but no decision has actually been made.
4
Post-lunch approvals
Hiring approvals, promo committee votes, PIP sign-offs made in a groggy 2pm slot correlate with more defaults and fewer overrides.
5
Anywhere calendar tetris has stacked judgement work
The signal is simple — if a decision matters, don't schedule it as the eighth thing on someone's day.

Design fixes that actually work

Depleted-manager defaults vs designed decisions

Under fatigue, managers…

Pick the default (middle rating, status quo)
Use fewer words, weaker evidence
Choose 'no' when justification is easier
Rubber-stamp the loudest voice in the room
Confabulate rationales after the fact

Design counter-moves

Schedule high-stakes decisions in the first two hours of the day
Cap 'judgement decisions' per day to ~5
Require pre-committed written positions
Force independent scoring before group discussion
Split write-ups and decisions across days

Operating rules for HR and managers

1
Fresh-brain calibration
Never run calibration on a Friday afternoon at the end of review week. Book it Tuesday–Thursday mornings, cap at 2 hours, one team per session.
2
Interview slot equity
Rotate which candidates are in the last slot. Better yet, standardise scoring so slot order matters less. Or split panels across days for senior loops.
3
Pre-commitment for high-stakes calls
Managers writing PIPs, promos, or terminations should draft a written position before the meeting where the decision is made. Fatigue-driven agreement in the room is a common failure mode.
4
Structured 1:1 templates
The 3–4 report of the day gets the same quality attention as the first when the template forces the same questions.
5
Decision hygiene metrics
Track: what proportion of hire/no-hire calls happen after 3pm? What proportion of promotion decisions happen in the last 20% of the cycle? These are leading indicators of quality drift.

The honest version

Willpower may or may not be a finite battery in the strict sense — but calendars, sleep, and cognitive load are real. Design your review, hiring, and 1:1 systems around a manager who is a normal human at 3pm on Thursday, not an idealised decision-maker at 9am on Monday.

FAQ

Frequently asked questions

Wait — is ego depletion real or not?

The strong version (any effortful task depletes willpower for any later task) is contested and largely didn't replicate in 2016. The narrower version — decision fatigue and cognitive load affecting judgement — is robust and directly relevant to HR.

So do I need to redesign every process?

No. Start with the highest-stakes decisions (hiring, promotion, PIP) and the visibly worst-scheduled ones (Friday afternoon calibration, back-to-back interview days). The biggest wins come from moving these to fresh-brain windows.

What about the manager who says 'I'm fine at 6pm'?

They're a normal human. The literature shows self-assessed fatigue correlates poorly with actual decision quality. Design around measured behaviour, not perceived stamina.

Takeaways

The strong 'willpower is a battery' claim didn't replicate — but decision fatigue in judgement work did.
Late-day, late-week, late-cycle decisions drift toward defaults. Design against that, not around it.
The highest-leverage fix is scheduling: move calibration, PIP calls, and final interviews into fresh-brain windows.
Independent written pre-commitments beat live-room deliberation for high-stakes calls.