Ego depletion and the willpower battery

In 1998 Roy Baumeister, Ellen Bratslavsky, Mark Muraven and Dianne Tice published an experiment that would be recounted in hundreds of subsequent papers and thousands of secondary accounts. Participants were seated in a room containing freshly baked chocolate chip cookies and a bowl of radishes. Some were instructed to eat the radishes and leave the cookies; others faced no such restriction. Everyone was then given geometric puzzles that had no solution, and the measure was how long each group persisted before giving up.

The radish group quit sooner. Baumeister's interpretation was that self-control draws on a shared and exhaustible resource: exert it on one task and less remains for the next. The metaphor of a battery followed, then the term ego depletion, then a research programme that ran for close to two decades.

The glucose extension

Matthew Gailliot and Baumeister later proposed a physiological substrate, arguing that the resource was blood glucose and reporting that sugared drinks restored performance after a depleting task. The claim drew sustained criticism on metabolic grounds. Robert Kurzban and others argued that the brain's energy consumption does not vary between tasks anywhere near enough to account for the reported effects, and subsequent work found that merely rinsing the mouth with a sugar solution without swallowing produced comparable results — which points to a signalling or motivational mechanism rather than fuel delivery.

The registered replication

A 2010 meta-analysis by Martin Hagger, Nikos Chatzisarantis and colleagues had reported a moderate overall effect, and the finding was widely treated as settled. In 2014, Evan Carter and Michael McCullough re-examined that same literature with methods designed to detect publication bias and reported that once small-study effects were modelled, the estimate shrank substantially — in some specifications to nothing.

The decisive test was a Registered Replication Report published in Perspectives on Psychological Science in 2016. Hagger and Chatzisarantis coordinated twenty-three laboratories running a single pre-agreed protocol on roughly two thousand participants, with the analysis plan fixed before data collection. The pooled effect was indistinguishable from zero.

Baumeister disputed the outcome, arguing that the committee had selected a depletion task poorly suited to producing the effect and that his laboratory had raised objections to the protocol before the study ran. That exchange is published and can be read alongside the report. A second multi-laboratory replication led by other groups subsequently reported similarly small estimates.

What survived

One line of work moved the explanation away from resources entirely. Veronika Job, Carol Dweck and Gregory Walton reported that depletion-like patterns appeared mainly among participants who held the belief that willpower is a limited resource, and were weak or absent among those who did not. On that account the original finding is not about fuel at all but about expectation — which reverses the causal story rather than rescuing it.

Michael Inzlicht and colleagues have proposed a different reframing, treating the pattern as a shift in motivation and attention over time rather than the consumption of a substance. Inzlicht, who worked within the depletion paradigm for years, has written publicly about revising his own position.

Cairn Post marks this entry as contested. The original experiments were conducted and published; the effect has not survived pre-registered attempts to reproduce it at the size originally reported, and the dispute over why remains open in the literature.