Most of what the social mind does happens between people — conformity, obedience, the pull of the group, each covered elsewhere in this series. But one question sits a little differently. It's not about other people reshaping you. It's about whether the act of resisting — of controlling yourself — uses up something finite. Whether willpower is a resource. That question produced one of the most influential findings in 20th-century psychology, and one of the most embarrassing collapses in 21st-century science.

In 1998, Roy Baumeister (b. 1953, United States) — a social psychologist at Case Western Reserve University — published a study that seemed to confirm what everyone already suspected. Participants who had to resist eating chocolate chip cookies sitting right in front of them gave up faster on a subsequent frustrating puzzle than participants who hadn't been asked to resist anything. Baumeister's explanation: self-control draws on a limited mental resource. Use it up on one task, and you have less for the next. He called it ego depletion. The idea spread fast. By the 2010s, it had generated over a thousand follow-up studies and was cited as evidence for everything from why judges give harsher sentences before lunch to why you shouldn't grocery shop hungry.

Position A — ego depletion is real — still has serious defenders, and their case isn't trivial. The original findings replicated repeatedly throughout the 2000s. Baumeister and his colleagues point out that the 2016 multi-lab study used a specific protocol — one that may have been too weak to produce depletion in the first place. Meta-analyses pooling hundreds of studies still find a small but detectable effect. And there's a plausible mechanism: glucose. Some research suggests that acts of self-control temporarily lower blood glucose in the brain, and that consuming sugar can restore performance. The strongest version of this position isn't 'willpower is a muscle that gets tired' — it's 'self-control is a biological process with real metabolic costs, and those costs accumulate.'

Position B — ego depletion is mostly an artifact — is now the majority view among researchers who study replication, and their evidence is pointed. The 2016 Many Labs study wasn't the only failure. A 2010 meta-analysis found signs of publication bias so severe that the average effect size in the literature was probably inflated by a factor of two or more. The glucose hypothesis has been directly tested and largely failed — drinking a non-caloric sweet drink (which shouldn't change blood glucose) produced the same 'restoration' effect as actual sugar, suggesting the effect was expectation, not metabolism. The deepest critique is structural: most ego depletion studies were small, run by researchers who believed the hypothesis, and published in journals that preferred positive results. That's a recipe for false positives regardless of what's actually true.