The Radish Experiment
In 1998, Roy Baumeister, Ellen Bratslavsky, Mark Muraven, and Dianne Tice ran one of the most influential — and, as it turned out, most contested — experiments in social psychology. They invited undergraduate participants to a laboratory that smelled of freshly baked chocolate chip cookies. On a table sat two plates: one heaped with those warm cookies and chocolate candies, the other with raw radishes. Participants were assigned to one of three conditions. The "cookie group" could eat freely from the cookies and chocolates. The "radish group" was told they could only eat from the radish plate — they had to actively resist the far more tempting items that were sitting right there in front of them. A control group received no food at all.
After this manipulation, all participants were given a geometric puzzle to work on. What they were not told: the puzzle was unsolvable. The researchers were measuring how long participants persisted before giving up.
The results were stark. The cookie group and the no-food control group persisted for roughly 20 minutes each. The radish group — those who had spent their time resisting the cookies — gave up after an average of just 8 minutes. They made fewer attempts, showed more signs of frustration, and quit far sooner.
The interpretation Baumeister and colleagues offered was simple and arresting: resisting the cookies consumed something. Whatever internal resource the radish participants had used to override their impulse to eat the forbidden food, that resource was now depleted, leaving them with less staying power when they encountered a subsequent frustrating task. They called this phenomenon ego depletion.
The paper, published in the Journal of Personality and Social Psychology, has been cited more than 5,000 times. It launched a subfield. It also became, two decades later, a central exhibit in psychology's reckoning with its own methods.
Definition
Ego depletion is the hypothesis that acts of self-regulation — effortful override of impulses, desires, or habitual responses — draw upon a limited internal resource, such that performing one act of self-control reduces the capacity for subsequent acts of self-control in the short term.
Ego Depletion vs. Motivated Performance
| Dimension | Ego Depletion | Motivated Performance |
|---|---|---|
| Resource assumption | Self-control draws on a finite, depletable store | Performance is governed by motivation, which can be recruited on demand |
| Effect of prior effort | Prior self-control tasks reduce subsequent performance | Prior effort can be offset by incentive, meaning, or importance |
| Mechanism proposed | Metabolic or cognitive resource consumption | Goal salience, attention allocation, effort calibration |
| Reversible by? | Rest, glucose replenishment, positive affect, or time | Shifting motivation, reframing stakes, social accountability |
| Key evidence | Cross-task performance decrements after self-control | Studies showing depletion effects disappear with monetary incentives or emergency framing |
| Theoretical lineage | Psychoanalytic ego, strength models, resource models | Expectancy-value theory, self-determination theory |
| Replication status | Contested; large preregistered replication failed | Generally robust; motivation effects replicate consistently |
Cognitive Science Foundations
The Strength Model of Self-Control
The intellectual core of ego depletion is what Baumeister and colleagues came to call the "strength model" of self-control. The model draws an analogy to muscle fatigue: just as a muscle exerts force by drawing on finite energy stores, the self exerts regulatory control by drawing on some central resource. Use the resource, and it depletes; rest, and it recovers.
Mark Muraven and Roy Baumeister formalized this model in a 2000 review article in Psychological Bulletin, surveying evidence that self-regulatory failures cluster — that people who fail at one form of self-control tend simultaneously to fail at others, and that prior regulatory demands consistently undermined subsequent ones across a wide range of behaviors, from physical endurance to cognitive suppression to decision-making.
Matthew Gailliot and Baumeister extended this framework in a 2007 paper in Psychological Review, proposing that the depleted resource was, literally, blood glucose. Their argument: the brain is a metabolic organ; self-regulation is cognitively demanding; demanding cognition consumes glucose; therefore self-control should be impaired when blood glucose is low and restored when glucose is replenished. They reported evidence that acts of self-regulation lowered blood glucose levels in participants, and that drinking a glucose-sweetened beverage (as opposed to an artificially sweetened one) restored self-control performance.
This glucose hypothesis was appealing in its concreteness. It also attracted pointed criticism. Roy Baumeister and colleagues had measured blood glucose changes in the range of 1 to 2 milligrams per deciliter — changes far too small to plausibly affect brain function in a healthy adult, whose brain normally operates within a glucose range of roughly 70 to 100 mg/dL and is protected by tight homeostatic regulation. Gailliot and Baumeister's 2007 paper was subsequently subject to several failed replications, including a direct replication attempt by Lange and Eggert published in Appetite in 2014, which found no glucose effect on self-control under controlled conditions.
Process Model and the Motivational Alternative
Michael Inzlicht and Brandon Schmeichel published a significant theoretical challenge in 2012 in Perspectives on Psychological Science. Their "process model" of ego depletion proposed that the performance decrements observed after self-regulatory effort were not caused by resource exhaustion in any metabolic sense. Instead, depleted participants shifted their motivational orientation: having exerted effort, they became more attentive to rewarding stimuli and less willing to continue investing in effortful tasks. Depletion, on this account, is not a tank running empty — it is a recalibration of effort allocation. The process model predicted that depletion effects could be reversed or eliminated not just by replenishing some hypothetical resource, but by changing what participants wanted to do or what they believed was at stake.
This reframing had significant empirical support. Carol Dweck, Veronika Job, and Gregory Walton published a striking study in Psychological Science in 2010 demonstrating that whether a person showed ego depletion depended substantially on whether they believed in it. Participants who held the intuitive view that mental effort is exhausting showed the classic depletion pattern. Participants who held the view that engaging in a demanding mental task can be energizing or has no cost to subsequent tasks showed no depletion effect — they performed equivalently on a second task regardless of how depleting the first task was. The belief, not just the behavior, appeared to moderate the outcome.
This finding had profound implications. If depletion is partly constituted by expectation and belief, then the phenomenon is not a straightforward consequence of burning through a biological resource. It becomes entangled with motivation, framing, and meaning — none of which fit comfortably inside a strength model.
Four Case Studies Across Domains
Case Study 1: Parole Decisions and the Judicial Depleted State
The most widely cited real-world application of ego depletion outside the laboratory comes from a 2011 paper by Shai Danziger, Jonathan Levav, and Liora Avnaim-Pessoa, published in the Proceedings of the National Academy of Sciences. The researchers analyzed 1,112 parole board hearings conducted by eight Israeli judges over a ten-month period.
They found a striking temporal pattern: the proportion of favorable rulings (granting parole) was approximately 65 percent at the start of each session — morning or post-lunch break — and fell sharply toward zero by the end of the session. After a food break, the rate reset to roughly 65 percent again. The authors interpreted this pattern as consistent with ego depletion: as judges exhausted their mental resources over the course of a session, they defaulted to the cognitively easier and less consequential choice — denial — rather than investing effort in individualized case evaluation.
The study received enormous media coverage and became canonical in discussions of decision fatigue. However, subsequent reanalysis raised questions. Andreas Glockner published a commentary in 2016 in PLOS ONE arguing that the pattern could be explained by case ordering (attorneys may strategically schedule their best cases at the start of sessions), and that the researchers had not adequately controlled for the difficulty of cases considered at different time points. The parole study, like the radish experiment, illustrates both the intuitive appeal and the methodological complexity of the ego depletion framework.
Case Study 2: Dietary Self-Regulation and Consumer Behavior
Kathleen Vohs and Ronald Faber published research in 2007 in the Journal of Consumer Research examining how prior acts of self-regulation affected spending behavior. Participants who had been asked to engage in an attention-control task (watching a silent video clip of a woman talking while a series of words appeared on screen, and suppressing the natural tendency to read the words) subsequently showed greater impulsive buying tendencies. They also spent more real money when given the opportunity to purchase items.
The implication for consumer psychology: exposure to decision-demanding environments — the sheer volume of choices required during a shopping trip, for instance — may compromise the self-regulatory capacity needed to resist impulse purchases later. Retailers may implicitly exploit this dynamic through store layouts that front-load consideration-heavy decisions, eroding self-control by the time customers reach checkout impulse displays.
Case Study 3: Aggression and Intimate Partner Behavior
Brad Bushman, C. Nathan DeWall, Richard Pond, and Michael Hanus published a study in 2014 in the Proceedings of the National Academy of Sciences examining how glucose levels (as a proxy for self-regulatory resources) related to partner-directed aggression in married couples. Couples who had lower blood glucose at baseline showed higher rates of aggressive behavior toward each other as measured by pins stuck into a voodoo doll and blasts of loud noise administered to partners in a laboratory game.
While this study used glucose as a marker rather than directly manipulating depletion, it extended the ego depletion framework into relationship dynamics, suggesting that the daily rhythms of hunger and satiety might influence the quality of interpersonal regulation — a claim that, if robust, would have substantial implications for conflict timing, intervention design, and even couples counseling scheduling.
Case Study 4: Ethical Decision-Making and Workplace Integrity
Christopher Barnes, Lorenzo Lucianetti, Dishan Kamdar, and Michael Christian published a study in the Academy of Management Journal in 2015 examining how sleep deprivation — which they framed as compromising self-regulatory capacity — affected deviant workplace behavior. Employees who reported poorer sleep quality were rated by supervisors as engaging in more unethical behavior across the working day, with effects particularly pronounced later in the day when cumulative depletion would be greatest. The study used experience sampling methodology, capturing ratings at multiple time points, which allowed the researchers to examine within-person variation across time.
The applied implication is significant: organizational contexts that impose heavy cognitive demands or long hours without recovery may inadvertently erode the ethical self-regulation of their employees — not through bad character, but through the temporal dynamics of resource depletion.
Intellectual Lineage
The concept of ego depletion did not emerge from nowhere. Baumeister's framing explicitly drew on Sigmund Freud's construct of the ego as an executive agency that mediates between id impulses and superego prohibitions, and that expends psychic energy in doing so. Freud's hydraulic model of mental energy — where desire and defense operate like pressure in a closed system — anticipated the strength model in important structural ways, even if the specific mechanisms Baumeister proposed were empirical rather than metapsychological.
William James, writing in The Principles of Psychology in 1890, had noted that voluntary attention is inherently effortful and that sustained effort appears to exhaust something. He called willpower a "faculty" and observed that it seemed to tire with use — a proto-strength model articulated before behaviorism made such constructs unfashionable.
The behavioral economics tradition contributed the concept of decision fatigue, closely related to ego depletion. Amos Tversky and Daniel Kahneman's dual-process framework — System 1 (fast, automatic) and System 2 (slow, deliberate) — provided a cognitive architecture within which depletion could be understood: effortful System 2 processing is resource-demanding, and when resources are low, cognition defaults toward System 1 heuristics. Kahneman's Thinking, Fast and Slow (2011) prominently featured Danziger and colleagues' Israeli judges study, helping to bring ego depletion into mainstream public discourse.
Within social psychology, Albert Bandura's concept of self-efficacy and Walter Mischel's studies of delay of gratification — famously, the Stanford marshmallow experiments — provided parallel empirical frameworks for studying self-regulation, though without the specific resource-depletion claim. Mischel's later reanalysis, published with Yuichi Shoda and Philip Peake in 1990 in Psychological Review, showed that young children's ability to delay gratification predicted SAT scores and adult adjustment, reinforcing the view that self-control is a stable individual capacity — one that, within the depletion framework, could also be temporarily impaired by contextual demands.
Empirical Research and the Replication Crisis
For roughly fifteen years after 1998, ego depletion generated an accumulating body of apparently supportive evidence. A 2010 meta-analysis by Hagger, Wood, Stiff, and Chatzisarantis published in Psychological Bulletin synthesized 83 studies involving over 200 experiments and reported a mean effect size of d = 0.62 — a medium-to-large effect by conventional standards. The analysis covered a wide range of depletion tasks (Stroop tasks, emotional suppression, attention control, resisting food) and outcome measures (persistence, performance, physical endurance), finding consistent evidence of cross-task self-control decrements.
This meta-analysis seemed to confirm ego depletion as one of the more robustly established phenomena in social psychology. Then came the replication crisis.
The Hagger et al. 2016 Preregistered Replication
In 2016, Martin Hagger — lead author of the supportive 2010 meta-analysis — led a massive preregistered multi-site replication of ego depletion. The study, published in Perspectives on Psychological Science, involved 23 laboratories across North America, Europe, and Australia, testing 2,141 participants using a closely standardized protocol based on procedures that had reliably produced depletion effects in the original literature.
The result: the overall effect size was d = 0.04, with a 95 percent confidence interval that included zero. The replication found essentially no evidence that the depletion manipulation produced worse performance on subsequent tasks. This was not a modest reduction from the original d = 0.62 — it was a near-complete absence of the effect.
Hagger and colleagues were careful in their interpretation. They did not conclude that ego depletion does not exist. Rather, they argued that the original published literature likely suffered from publication bias (studies showing no depletion effect were not published), that many original studies were underpowered and thus vulnerable to false positives, and that effect sizes in the published literature were inflated by selective reporting.
The failure to replicate ignited substantial controversy. Baumeister himself disputed aspects of the replication protocol, arguing that the specific depletion task used (crossing out the letter "e" under attention-control rules) was not an established paradigm with a track record of producing depletion, and that laboratories with experimenter-demand characteristics different from the original studies might have diluted the effect. Other researchers, including Inzlicht, argued that the replication failure was expected precisely because the original effect was fragile, situationally contingent, and moderated by belief — consistent with the process model critique.
Prior Publication Bias
A funnel plot analysis of the 2010 meta-analysis data conducted by Carter and McCullough in 2013 and published in Frontiers in Psychology had already raised red flags before the 2016 preregistered replication. Their analysis suggested substantial asymmetry in the distribution of published effect sizes — a pattern consistent with the selective non-publication of null results. If the published literature overrepresented significant, positive findings, then the apparent robustness of ego depletion may have reflected the sociology of science as much as the psychology of self-control.
This problem is not unique to ego depletion. Across social psychology, the replication crisis has revealed that many phenomena with apparently large, consistent effect sizes in the published literature fail to replicate when tested with larger, pre-registered designs. The ego depletion case is notable primarily because the original meta-analysis was unusually comprehensive and the subsequent disconfirming replication unusually large and well-organized.
Limits and Nuances
The Belief Moderator
Veronika Job, Carol Dweck, and Gregory Walton's 2010 findings — that beliefs about the nature of willpower moderate the depletion effect — have been replicated multiple times, including in Bertrams and Baumeister's 2015 work. This creates a genuinely strange situation for the resource model: if the effect depends on whether you believe in the effect, it is difficult to maintain that what is being depleted is a biological resource indifferent to belief. Resources like blood glucose do not replenish themselves because you believe they have.
One reconciliation: the resource model and the process model may be compatible if the resource is not metabolic energy per se but something more like motivational engagement — a cognitive construct that is, by its nature, susceptible to framing and belief. But this reconciliation comes at the cost of the model's clean empirical clarity.
Trait Self-Control Matters
Muraven, Shmueli, and Burkley demonstrated in 2006 in the Journal of Personality and Social Psychology that people high in trait self-control — who generally regulate more effectively across life domains — showed diminished depletion effects. High self-control individuals may be more practiced at the kinds of tasks used to induce depletion, and thus find them less demanding. Alternatively, they may have more capacity to spare. Either way, ego depletion is not equally likely across individuals.
Recovery Mechanisms Are Multiple and Contested
The original strength model suggested glucose as the recovery mechanism. That proposal has largely failed empirical scrutiny. Other proposed recovery mechanisms include positive affect (Tice, Baumeister, Shmueli, & Muraven, 2007, Journal of Experimental Social Psychology), which restored self-control capacity after depletion — apparently by motivationally reinvigorating participants rather than metabolically replenishing them. Sleep is a robust predictor of next-day self-control capacity. Mindfulness practice has been associated with reduced susceptibility to depletion effects in several studies.
Demand Characteristics and the Task Problem
A persistent methodological concern is that participants in depletion studies may infer from the experimental context that they are supposed to perform worse on the second task. If demand characteristics — the cues participants use to guess what the experimenter expects — drive performance, then depletion effects could be an artifact of participant compliance rather than genuine resource exhaustion. Pre-registration and double-blind designs mitigate this, but the 2016 multi-site replication used such controls and still found no effect, which cuts in multiple directions simultaneously.
What, If Anything, Is Being Depleted?
The fundamental unresolved question is mechanistic. The glucose model has not held up. The process model offers an alternative framing but does not specify a precise mechanism. More recent proposals invoke attention, motivation, and affect as the mediating variables. Some researchers, including Inzlicht and colleagues writing in Trends in Cognitive Sciences in 2014, have proposed abandoning the resource metaphor entirely in favor of a signal-detection or opportunity-cost account: organisms continuously monitor the costs and benefits of cognitive effort, and what appears to be depletion is actually a rational downregulation of effort when the perceived costs exceed the perceived benefits.
This shift, if correct, would be a significant conceptual revision. It would mean that ego depletion, as originally framed, is wrong — but that the observations motivating it (people do seem to perform worse after extended self-regulatory effort) capture something real that demands explanation. The explanation, however, would belong to motivational economics rather than metabolic resource management.
What This Means in Practice
The ego depletion literature, even in its contested state, has generated practical insights that remain worth taking seriously, with appropriate calibration.
The evidence that self-regulatory demands cluster — that people who are stressed, sleep-deprived, and making too many decisions are more likely to fail at additional acts of regulation — is broadly consistent across methodologies even if the specific resource model is uncertain. Whether the mechanism is metabolic, motivational, or attentional, structuring environments to reduce unnecessary self-regulatory demands appears protective.
The Israeli judges data, contested as it is, points toward the value of scheduled breaks in cognitively demanding roles. Whether the mechanism is glucose restoration, attentional refresh, or simply a reset in motivational orientation, the practical takeaway is robust: judges who take breaks make different decisions, and those decisions are more favorable to defendants. The mechanism matters for theory; the pattern matters for policy.
Job, Dweck, and Walton's belief findings suggest that cultivating what they called a "nonlimited" theory of willpower — treating engagement in effortful tasks as potentially energizing rather than inevitably depleting — may reduce susceptibility to depletion effects. This is not an invitation to ignore limits, but it suggests that framing and expectation are not epiphenomenal. How you think about effort may shape how much of it you have available.
Finally, the replication crisis surrounding ego depletion is itself a substantive finding about scientific practice: that intuitive, theoretically coherent phenomena can accumulate thousands of citations and meta-analytic confirmation before being subjected to adequately powered, preregistered tests. The ego depletion story is now as much a lesson in research methodology as it is a lesson in self-control.
References
Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998). Ego depletion: Is the active self a limited resource? Journal of Personality and Social Psychology, 74(5), 1252–1265.
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of limited resources: Does self-control resemble a muscle? Psychological Bulletin, 126(2), 247–259.
Gailliot, M. T., & Baumeister, R. F. (2007). The physiology of willpower: Linking blood glucose to self-control. Personality and Social Psychology Review, 11(4), 303–327.
Hagger, M. S., Wood, C., Stiff, C., & Chatzisarantis, N. L. D. (2010). Ego depletion and the strength model of self-control: A meta-analysis. Psychological Bulletin, 136(4), 495–525.
Danziger, S., Levav, J., & Avnaim-Pessoa, L. (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences, 108(17), 6889–6892.
Job, V., Dweck, C. S., & Walton, G. M. (2010). Ego depletion — Is it all in your head? Implicit theories about willpower affect self-regulation. Psychological Science, 21(11), 1686–1693.
Inzlicht, M., & Schmeichel, B. J. (2012). What is ego depletion? Toward a mechanistic revision of the resource model of self-control. Perspectives on Psychological Science, 7(5), 450–463.
Hagger, M. S., Chatzisarantis, N. L. D., et al. (2016). A multilab preregistered replication of the ego-depletion effect. Perspectives on Psychological Science, 11(4), 546–573.
Carter, E. C., & McCullough, M. E. (2013). Publication bias and the limited strength model of self-control: Has the evidence for ego depletion been overestimated? Frontiers in Psychology, 4, 823.
Vohs, K. D., & Faber, R. J. (2007). Spent resources: Self-regulatory resource availability affects impulse buying. Journal of Consumer Research, 33(4), 537–547.
Inzlicht, M., Schmeichel, B. J., & Macrae, C. N. (2014). Why self-control seems (but may not be) limited. Trends in Cognitive Sciences, 18(3), 127–133.
Muraven, M., Shmueli, D., & Burkley, E. (2006). Conserving self-control strength. Journal of Personality and Social Psychology, 91(3), 524–537.
Frequently Asked Questions
What is ego depletion?
Ego depletion is the hypothesis that self-control draws on a limited psychological resource — analogous to a muscle — that becomes depleted with use, resulting in reduced capacity for subsequent acts of self-control. Roy Baumeister, Ellen Bratslavsky, Mark Muraven, and Dianne Tice introduced the concept in their 1998 Journal of Personality and Social Psychology paper, showing that subjects who had exerted self-control on one task subsequently performed worse on an unrelated self-control task. The theory proposed that diverse forms of self-regulation — resisting temptation, suppressing emotions, making decisions — draw on the same limited resource.
What did the radish and cookie experiment find?
Baumeister et al. (1998) placed subjects in a room with a plate of freshly baked cookies and a plate of radishes. Some subjects were told to eat only radishes (depleting their self-control by resisting the cookies); others ate cookies freely; a control group had no food. All subjects then worked on an unsolvable geometric puzzle, with persistence measured. Radish-group subjects quit after an average of 8 minutes. Cookie and control groups persisted for approximately 20 minutes. The act of resisting temptation had measurably consumed a resource that the subjects needed for subsequent persistence.
Did ego depletion replicate?
The replication record is mixed and contested. Hagger et al.'s 2010 meta-analysis of 83 studies found a mean effect size of d = 0.62. However, Hagger et al.'s 2016 pre-registered multi-lab replication — 23 independent laboratories, 2,141 participants — found an effect of d = 0.04, statistically indistinguishable from zero. Carter and McCullough's 2013 funnel-plot analysis found evidence of publication bias in the original literature. Baumeister and colleagues objected that the replication protocol was inadequate. The controversy remains unresolved, but the consensus has shifted toward viewing ego depletion as a smaller, more conditional effect than the original literature suggested.
What did the Israeli judges study show?
Danziger, Levav, and Avnaim-Pessoa's 2011 PNAS analysis of 1,112 parole board decisions by eight Israeli judges found that the probability of a favorable ruling started at approximately 65% after a food break, then declined sharply to near zero as the session progressed, then reset to 65% after the next break. The authors attributed the pattern to ego depletion: as judges made more decisions, their capacity for effortful, individualized judgment depleted, causing them to default to the easier denial ruling. The study became widely cited. Subsequent reanalysis has challenged the ego depletion interpretation, suggesting the pattern may reflect other factors including ordering effects and meal timing.
Does believing in willpower affect whether ego depletion occurs?
Veronika Job, Carol Dweck, and Gregory Walton's 2010 Psychological Science study found that subjects who held an implicit theory that willpower is unlimited — that exerting self-control does not deplete it — showed no ego depletion effects across multiple tasks. Subjects who held a limited-resource theory showed standard depletion effects. The finding suggests that ego depletion may be partly self-fulfilling: if you believe willpower depletes, it does; if you believe it does not, it does not. This does not resolve the mechanism debate but adds an important moderator that earlier research had not identified.