It has been claimed that the psychological literature is filled with zombie theories—walking dead that are not supported by evidence, no longer believed by insiders or even by the original proponents, but that live on in mindless citations and textbooks forever. Sometimes, though, they do die in silence, without a funeral or obituary. One example of a dead theory in psychology is the glucose theory of willpower.
While the inventor of this theory (and the underlying phenomenon), Roy F. Baumeister, still clings to his broader theory of willpower despite two large replication failures, even he has walked away from the idea that willpower depends on blood-glucose levels.
1. Gailliot’s Original Glucose Studies (2007–2009)
Baumeister and Gailliot’s studies (e.g., Gailliot et al., 2007, Journal of Personality and Social Psychology; Gailliot & Baumeister, 2007, Psychological Science) claimed that exerting self-control reduces blood-glucose levels and that restoring glucose—even by drinking a sugary beverage—replenishes willpower and improves performance on subsequent self-control tasks.
The reported effects were dramatic. Across a series of small-sample experiments, the authors found large, consistently significant results, suggesting a robust physiological mechanism underlying self-control. The simplicity of the claim—that willpower literally runs on sugar—made the theory intuitively appealing and easy to test. These findings immediately inspired a wave of replications and extensions, many of which were conceptual replications using similar small-sample, between-subjects designs.
2. Schimmack’s 2012 “Incredibility Index” Analysis
In 2012, Ulrich Schimmack applied his Incredibility Index (IC-index) to Gailliot’s published results—a meta-analytic tool that tests whether the reported proportion of significant results is credible given the observed power of the individual studies.
The results were striking. The distribution of p-values in Gailliot’s papers was too good to be true. The success rate (about 100%) was incompatible with the small sample sizes, and even with the inflated effect-size estimates, the estimated average power of these studies was far too low to justify only significant results.
During the review process, Roy F. Baumeister openly admitted that the published studies were selected from a larger set that included many null results. He justified this practice by claiming that this was simply what everyone did—an argument that, while true at the time, highlighted how pervasive questionable research practices had become in psychology. At that point, it was still unclear whether these practices merely inflated real effects or had created an entirely spurious phenomenon.
3. The Fallout: A Futile Wave of Follow-Up Research
Yet the glucose theory continued to attract attention—precisely because those early, inflated findings appeared compelling. For several years, researchers treated it as a promising physiological explanation for self-control. However, the replication crisis made it possible to publish replication failures, and several articles reported that they could not find effects of glucose on willpower.
From 2012 to 2018, dozens of researchers tried to replicate the glucose–willpower effect and found mostly null or inconsistent results. Eventually, large-scale meta-analyses—Vadillo, Gold, and Osman (2016) and Lange and Eggert (2014)—confirmed the suspicion: the literature was biased, the true effect size was near zero, and the original findings likely reflected p-hacking and selective publication.
4. The Current Status (Post-2020): The Glucose Model Is Dead
Today’s consensus—including Baumeister’s own 2024 review—effectively concedes that the glucose theory of willpower is dead, despite the initially strong-looking evidence for it. That evidence only appeared strong but was in fact illusory, based on unscientific practices that inflated effect sizes and suppressed null results.
Even the broader theory of ego-depletion has come under heavy criticism, as it was developed and supported using the same questionable practices. Two large replication studies failed to reproduce the basic ego-depletion effect and produced effect-size estimates close to zero. Baumeister continues to defend the broader theory, so it cannot yet be declared dead, but the glucose model has vanished.
5. Why This Matters for Meta-Science
The glucose theory of willpower is more than just a failed idea—it is a vivid example of how questionable research practices can create the illusion of discovery and lead to years of wasted effort. Schimmack’s 2012 incredibility analysis exposed the statistical impossibility of Gailliot’s findings long before the replication failures confirmed it.
6. Postscript
The future will show how many other walking-dead theories and ghostly phenomena, created in laboratories through dark research practices, will eventually die and join the glucose theory in the graveyard of psychological ideas that lack empirical support. To make this happen, psychology needs more courageous scientists—modern-day Dr. Van Helsings—who are willing to put a wooden stake through the heart of theories that deserve to die.
Happy Halloween
What’s incredible to me is that you sounded the alarm with your Incredibility Index back in 2012. Yet it took 12 years (12 years!) for Baumeister to stake his resource-sucking willpower depends on glucose theory. That’s a long time.
And the scary part is, as you indicate, that there may be many such theories that trickle down into public policy, classrooms, and everyday parlance and persist as modern-day ghosts.
What is most scary to me is how much psychologists resist to examine the foundations of their belief system. More a cult than a science. 🙁