The willpower battery that didn't replicate

For years psychology taught that self-control runs down like a battery. Then dozens of labs ran the same test the same way, and the effect all but vanished.

In 1998 Roy Baumeister and his colleagues published an experiment that seemed to explain a great deal about everyday life. People who had been asked to resist tempting chocolates later gave up sooner on a difficult, frustrating puzzle. Holding back once seemed to leave less in reserve for the next effort.

The idea became known as ego depletion: self-control draws on a limited resource, and using some of it makes the next act of self-control harder. It was intuitive, and it matched how a long day feels.

Many studies followed. A 2010 meta-analysis by Martin Hagger and colleagues combined 198 independent tests and found a moderate effect, a d of about 0.6, and judged it extremely unlikely that the effect wasn't real.

Then came the checks

A meta-analysis can only combine what gets published, and studies that find nothing are less likely to be published. When Evan Carter and Michael McCullough corrected for that bias across more than 100 studies, the estimate fell to about 0.2, not significantly different from zero.

The clearest answer came from a different kind of study. In 2016, 23 labs ran one agreed procedure with 2,141 participants in total, with the method fixed before the data came in. They found no evidence of ego depletion. In 2021 a second multisite project, led by Kathleen Vohs, ran 36 labs and 3,531 participants and found an effect of 0.06, about a tenth of the 2010 estimate, and again no evidence of depletion.

The d here is a standard way to say how far apart two groups are, measured against how much people vary within them. At 0.06, any difference between people who had just exercised self-control and people who hadn't is far too small to notice.

None of this shows that tiredness never affects what we do, or that self-control is unlimited. The conclusion is narrower: the specific lab effect, one short self-control task weakening the next, doesn't reliably appear when many labs test it the same way.

The replication crisis in psychology is often told as an embarrassment. This is its other side. The field built a way to check its own findings, used it on a famous one, and changed its mind.