ArticlePDF Available


According to the strength model, self-control is a finite resource that determines capacity for effortful control over dominant responses and, once expended, leads to impaired self-control task performance, known as ego depletion. A meta-analysis of 83 studies tested the effect of ego depletion on task performance and related outcomes, alternative explanations and moderators of the effect, and additional strength model hypotheses. Results revealed a significant effect of ego depletion on self-control task performance. Significant effect sizes were found for ego depletion on effort, perceived difficulty, negative affect, subjective fatigue, and blood glucose levels. Small, nonsignificant effects were found for positive affect and self-efficacy. Moderator analyses indicated minimal variation in the effect across sphere of depleting and dependent task, frequently used depleting and dependent tasks, presentation of tasks as single or separate experiments, type of dependent measure and control condition task, and source laboratory. The effect size was moderated by depleting task duration, task presentation by the same or different experimenters, intertask interim period, dependent task complexity, and use of dependent tasks in the choice and volition and cognitive spheres. Motivational incentives, training on self-control tasks, and glucose supplementation promoted better self-control in ego-depleted samples. Expecting further acts of self-control exacerbated the effect. Findings provide preliminary support for the ego-depletion effect and strength model hypotheses. Support for motivation and fatigue as alternative explanations for ego depletion indicate a need to integrate the strength model with other theories. Findings provide impetus for future investigation testing additional hypotheses and mechanisms of the ego-depletion effect.
Ego Depletion and the Strength Model of Self-Control: A Meta-Analysis
Martin S. Hagger, Chantelle Wood, and Chris Stiff
University of Nottingham Nikos L. D. Chatzisarantis
National Institute of Education, Singapore
According to the strength model, self-control is a finite resource that determines capacity for effortful
control over dominant responses and, once expended, leads to impaired self-control task performance,
known as ego depletion. A meta-analysis of 83 studies tested the effect of ego depletion on task
performance and related outcomes, alternative explanations and moderators of the effect, and additional
strength model hypotheses. Results revealed a significant effect of ego depletion on self-control task
performance. Significant effect sizes were found for ego depletion on effort, perceived difficulty,
negative affect, subjective fatigue, and blood glucose levels. Small, nonsignificant effects were found for
positive affect and self-efficacy. Moderator analyses indicated minimal variation in the effect across
sphere of depleting and dependent task, frequently used depleting and dependent tasks, presentation of
tasks as single or separate experiments, type of dependent measure and control condition task, and source
laboratory. The effect size was moderated by depleting task duration, task presentation by the same or
different experimenters, intertask interim period, dependent task complexity, and use of dependent tasks
in the choice and volition and cognitive spheres. Motivational incentives, training on self-control tasks,
and glucose supplementation promoted better self-control in ego-depleted samples. Expecting further acts
of self-control exacerbated the effect. Findings provide preliminary support for the ego-depletion effect
and strength model hypotheses. Support for motivation and fatigue as alternative explanations for ego
depletion indicate a need to integrate the strength model with other theories. Findings provide impetus
for future investigation testing additional hypotheses and mechanisms of the ego-depletion effect.
Keywords: self-control strength, self-regulation, limited resource, dual-task paradigm, research synthesis
Supplemental materials:
In everyday life people show a remarkable capacity to regulate
the self and overcome the impulses and drives that tempt us to
overeat, drink too much alcohol, take harmful recreational drugs,
engage in violent actions when provoked, say hurtful things to
others, spend money beyond their means, engage in inappropriate
sexual activity, or procrastinate when they should be working
(Steel, 2007; Tice & Bratslavsky, 2000). This ability to attain
deliberative control over impulses (Ainslie, 1975; Eisenberg et al.,
2003; Fujita & Han, 2009) and abstain from gratifying immediate
needs and desires (Metcalfe & Mischel, 1999; Mischel, Shoda, &
Rodrieguez, 1989) is extremely adaptive and enables people to
engage in goal-directed behavior to bring about long-term desir-
able outcomes (Baumeister, 2005; Fishbach & Labroo, 2007;
Logue, 1988). If people were unable to regulate their behavior, life
would become a series of unconstrained impulsive actions to
service immediate urges, desires, and emotions. Goal-directed
behavior and the achievement of long-term outcomes would be-
come impossible, as people would not be able to engage in the
disciplined, focused effort required (Loewenstein, 1996).
Despite the human capacity to regulate the self, many behav-
ioral and social problems stem from persistent lapses of self-
control. Problems such as obesity, drug abuse, violent crime,
inability to manage finances (including personal debt and gam-
bling problems), unplanned pregnancy, eating disorders, sexu-
ally transmitted disease, and some chronic diseases (e.g., cancer
and heart disease) have their roots, directly or indirectly, in
self-regulation failure (Baumeister, Heatherton, & Tice, 1994;
Muraven & Baumeister, 2000; Wills & Stoolmiller, 2002).
Analogously, successful regulation of the self contributes to
many adaptive outcomes in society, such as success at school,
at college, and in the workplace; cohesive personal relation-
ships; superior physical and mental health; better ability to cope
with problems; and reduced susceptibility to social ills like drug
abuse and criminality (Gailliot & Baumeister, 2007a; Hammer,
2005; Levy, 2006; Tangney, Baumeister, & Boone, 2004). The
high importance of self-regulation to such behaviors and con-
comitant outcomes has meant that it has become the focus of a
considerable body of research in social psychology.
Martin S. Hagger, Chantelle Wood, and Chris Stiff, School of Psychol-
ogy, University of Nottingham, Nottingham, Nottinghamshire, United
Kingdom; Nikos L. D. Chatzisarantis, National Institute of Education,
Nanyang Technical University, Singapore.
Chantelle Wood is now at the Institute of Psychological Sciences,
University of Leeds, Leeds, West Yorkshire, United Kingdom. Chris Stiff
is now at the School of Psychology, Keele University, Keele, Staffordshire,
United Kingdom.
The research was supported in part by Leverhulme Trust Grant
F/00568/R, awarded to Nikos L. D. Chatzisarantis and Martin S. Hagger.
We thank James M. Tyler and Rex A. Wright for their helpful comments
on an earlier draft of this paper.
Correspondence concerning this article should be addressed to Martin S.
Hagger, Personality, Social Psychology, and Health Research Group, School
of Psychology, University of Nottingham, University Park, Nottingham, Not-
tinghamshire NG7 2RD, United Kingdom. E-mail: martin.hagger@
Psychological Bulletin © 2010 American Psychological Association
2010, Vol. 136, No. 4, 495–525 0033-2909/10/$12.00 DOI: 10.1037/a0019486
Research into self-regulation and failure of self-control has
frequently adopted cognitive models in which self-regulation is
viewed as a function of beliefs, judgments, expectations, attitudes,
and intentions (Ajzen, 1985; Bagozzi, 1992; Koestner, Bernieri, &
Zuckerman, 1992; Sansone & Smith, 2000). In contrast, capacity-
based approaches propose that self-control is a limited commodity
that restricts self-regulatory capability (Baumeister & Heatherton,
1996; Fishbach, Friedman, & Kruglanski, 2003; Metcalfe & Mis-
chel, 1999; Mischel et al., 1989). Inspired by this approach,
Baumeister and colleagues (Baumeister, Bratslavsky, Muraven, &
Tice, 1998; Baumeister & Heatherton, 1996; Baumeister, Vohs, &
Tice, 2007; Muraven & Baumeister, 2000; Vohs & Heatherton,
2000) developed the strength model of self-control.
A major tenet
of the model is that engaging in acts of self-control draws from a
limited “reservoir” of self-control that, when depleted, results in
reduced capacity for further self-regulation. In the model, self-
control is viewed as analogous to a muscle. Just as a muscle
requires strength and energy to exert force over a period of time,
acts that have high self-control demands also require strength and
energy to perform. Similarly, as muscles become fatigued after a
period of sustained exertion and have reduced capacity to exert
further force, self-control can also become depleted when demands
are made of self-control resources over a period of time. Baumeis-
ter and colleagues termed the state of diminished self-control
“strength” ego depletion.
An increasing body of research has supported the short-term
self-regulatory deficits predicted by the strength model (Baumeis-
ter, Gailliot, DeWall, & Oaten, 2006; Muraven & Baumeister,
2000). Self-control resource depletion has also been shown to
coincide with increased subjective and physiological effort, fa-
tigue, and task difficulty. The model has been further extended to
include hypotheses based on the strength or energy metaphor. For
example, anticipating future self-control demands results in people
conserving energy for future effort, just as an athlete spares his or
her muscles in anticipation of forthcoming demands (Muraven,
Shmueli, & Burkley, 2006). Regular training on tasks requiring
self-regulation can attenuate the ego-depletion effect; just as train-
ing a muscle increases its endurance and strength (Gailliot, Plant,
Butz, & Baumeister, 2007; Muraven, Baumeister, & Tice, 1999;
Oaten & Cheng, 2006a, 2006b, 2007). Finally, rest and recupera-
tion can regenerate self-control, just as a muscle’s strength returns
after a period of rest (Muraven & Baumeister, 2000; Tyler &
Burns, 2008).
Given the burgeoning literature in the field of self-control and
the strength model, a meta-analytic synthesis of the ego-depletion
effect is important and timely. This article makes a unique contri-
bution to knowledge on self-control by testing the size and con-
sistency of the ego-depletion effect in the extant literature. It
resolves whether inconsistencies in the size of the effect are due to
methodological artifacts or whether there is substantial variation in
the effect across studies due to the existence of extraneous mod-
erating variables. The analysis also tests whether the self-control
deficits observed in ego-depletion experiments can be accounted
for by alternative explanations—such as skill, fatigue, motivation,
self-efficacy, and negative affect—and whether these are consis-
tent with, or contradict, the strength model. Another aim was to
identify potential moderators of the effect, such as the defining
characteristics of self-control tasks and experimental design. The
effect of ego depletion on other variables that have been suggested
as indicators of depletion (e.g., effort, perceived difficulty, subjec-
tive fatigue, blood glucose levels) is also examined. Finally, the
conservation, training, and recovery hypotheses based on the
strength model are tested. The theoretical background to each of
these issues is discussed next.
Self-Control and the Strength Model
The strength model offers an explanation for self-control that
transcends cognitive and associative-learning models (Baumeister
et al., 1998). It predicts that acts of self-control draw from a
common, global resource. The resource is limited and vulnerable
to becoming depleted over time, just as a muscle becomes tired
after a period of exertion. As a consequence, after people have
engaged in an act of self-control, their capacity to exercise further
self-control becomes exhausted, leading to decreased performance
on subsequent acts of self-control. According to the model, once a
person’s self-control reserves have been depleted, the resulting
state of ego depletion can be counteracted by replenishing the
resource through rest or relaxation (Tyler & Burns, 2008) or by
taking on fuel (Gailliot, Baumeister, et al., 2007).
Empirical tests of the ego-depletion effect have typically
adopted an experimental procedure using two unrelated self-
control tasks, known as the dual-task paradigm (Baumeister et al.,
1998; Finkel et al., 2006; Muraven, Tice, & Baumeister, 1998).
Participants assigned to an experimental ego-depletion group are
required to engage in two consecutive tasks requiring self-control.
Control participants are also required to engage in two consecutive
tasks, but only the second task requires self-control. The strength
model predicts that experimental-group participants’ performance
on the second self-control task will be impaired relative to that of
control-group participants. This is because the finite self-control
resources of the experimental participants will be diminished after
the initial self-control task, leaving little to draw on for the second
task (Baumeister et al., 2007).
Adopting this paradigm, Baumeister and coworkers (Baumeister
et al., 1998; Muraven et al., 1998) provided initial support for the
ego-depletion effect. For example, one study required those in
the experimental group to suppress their emotions when watching
an evocative video; controls were informed they could freely
express their emotions. Participants were then required to hold a
spring-loaded handgrip to exhaustion. Compared with the control
group, participants who suppressed their emotions had signifi-
Self-control is defined as the capacity of the individual to alter, modify,
change, or override his or her impulses, desires, and habitual responses
(Baumeister & Heatherton, 1996). Quite literally, it is the ability of the self
to exert control over the self. Other terms often considered synonymous
with self-control include willpower, self-discipline, and self-regulation
(Duckworth & Seligman, 2005; Henden, 2008; Mischel, 1996). Self-
control and self-regulation, in particular, are often used interchangeably in
the literature (Baumeister et al., 2007; Hofmann et al., 2007). Self-control
can be viewed as a specific case of self-regulation in which the person
exerts deliberate and conscious effort to control the self, and self-regulation
is a global term that also encompasses reflexive and nonconscious regu-
latory processes such as homeostasis (Baumeister, Vohs, & Tice, 2007). In
this article we define self-control as the effortful capacity of the individual
to regulate his or her emotions, thoughts, impulses, or other well-learned or
automatic behavioral responses (Vohs, 2006).
cantly impaired performance on the handgrip task (Muraven et al.,
1998). In another study, participants assigned to the experimental
group were required to eat radishes and resist the temptation of
appetizing chocolates, but controls were asked to eat the choco-
lates and leave the radishes. The experimental participants exhib-
ited significantly lower persistence on a subsequent unsolvable
geometric puzzle task (Baumeister et al., 1998). According to the
strength model, suppressing emotions or resisting the tempting
food required the overriding of a habitual or dominant response
and led to the depletion of self-control resources. This resulted in
impaired performance on subsequent tasks due to the reduced
availability of self-control resources. The effect has been repli-
cated on numerous occasions by Baumeister and colleagues (for
reviews, see Baumeister & Vohs, 2007; Baumeister et al., 2007) as
well as researchers in other laboratories (e.g., Martijn et al., 2007;
Tyler, 2008; Wright, Martin, & Bland, 2003), and it has been
shown to be relatively consistent across different spheres or do-
mains of depletion. The findings provide support for a global
self-control resource, because the transfer of the effect across
spheres suggests that ego depletion is not an artifact of a particular
task or domain.
However, a minority of studies adopting these methods have
reported nonsignificant ego-depletion effects (e.g., Stillman, Tice,
Fincham, & Lambert, 2009; Wright et al., 2007; Wright, Stewart,
& Barnett, 2008). For example, Stillman et al. (2009) found that
participants assigned to an ego-depletion group that required them
to engage in a thought suppression task did not differ in their
persistence on a word production task relative to controls who
were not required to suppress their thoughts. Wright et al. (2007)
used a difficult counting task to deplete self-control resources and
found that performance on a subsequent mental arithmetic task did
not differ from that of a control group that performed an easy
initial counting task. These findings suggest that support for the
ego-depletion effect is not unequivocal, and there are variations
across the literature. Furthermore, the inconsistencies may be due
to the presence of moderating factors, such as the features of the
tasks used. The present meta-analytic synthesis of the findings of
ego-depletion studies therefore makes an important contribution to
the literature by estimating whether the effect is present in the
population, resolving inconsistencies in the literature, and testing
the degree of heterogeneity associated with the effect.
Alternative Explanations
Other explanations have been put forward to explain the self-
regulatory failures observed in ego-depletion experiments. The
aim in this section is to review these alternative explanations,
compare their tenets with those of the strength model, and identify
whether they are consistent with, compete with, or complement its
Self-control capacity can be conceptualized as a skill that is
developed over time and that enables people to actively invest the
required effortful action to bring about future goals or outcomes
(Carver & Scheier, 1998; Wills & Dishion, 2004). Baumeister and
coworkers (Baumeister et al., 1998; Muraven et al., 1998) pro-
posed that a skill model would predict relatively little change in
performance across the first and second tasks in dual-task para-
digm experiments, as people would merely apply the well-learned
skill to each task they encounter. According to this view, self-
control capacity is an acquired, relatively stable skill that, once
learned, would result in a relatively consistent performance on
consecutive self-control tasks. The consistent decrement in self-
control task performance observed in ego-depletion experiments
led Baumeister et al. (1998) to reject skill as a plausible explana-
However, Baumeister and colleagues recognized that people
may experience learning (Baumeister et al., 1998) or “warm-up”
(Muraven et al., 1998) periods in performance on self-control
tasks, particularly novel ones, suggesting that there may be cir-
cumstances in which performance improves with time. Indeed,
research in the area of skill acquisition and motor learning suggests
that people’s learning of skills over time is both gradual and
transferable (J. R. Anderson, 1982; Rosenbaum, Carlson, & Gil-
more, 2001). Therefore, in the early stages of the development of
self-control skills performance may increase but, assuming the
transferability of the self-control “skill,” this would eventually
stabilize and be applicable across different spheres or domains. In
addition, skill improvement is a relatively slow, long-term process,
but ego depletion has generally been tested as a relatively short-
term deficit in studies adopting the dual-task paradigm (Vohs et
al., 2008). As a result, short-term variations in self-control capacity
may not be influenced by the gradual changes in self-control
capacity due to the learning of self-control “skills.” This is sup-
ported by observed decrements in postdepletion second-task per-
formance in dual-task paradigm experiments rather than no change
or improvement. The overall weight of evidence seems to lend
support to a limited resource model of short-term self-control
capacity rather than a skill-based model.
Although much of the research on the strength model has
focused on short-term resource depletion (Vohs et al., 2008), there
is evidence that chronic experience or practice on self-control tasks
results in improvements in self-regulatory capacity. Individuals
practiced on self-control tasks appear to be less vulnerable to ego
depletion (Hui et al., 2009; Muraven et al., 1999). This is consis-
tent with the strength model in that exercise increases self-control
strength, allowing for a deeper pool of resources to be available for
subsequent performance on self-control tasks. Alternatively, a skill
model implies that the increased practice results in increases in the
efficiency with which self-control resources are used. This is not
inconsistent with the strength model, as a limited self-control
resource is implicated in the process, but it provides an alternative
to the “extended pool” explanation. A skill model may therefore
have utility in explaining long-term improvements in self-control
capacity. An aim in the present analysis was to examine the effects
of self-control practice as a means to promote self-regulatory
capacity. The issue of training and ego depletion is discussed
further in the section on extensions to the strength model.
Fatigue is likely to be implicated in the ego-depletion effect.
Theory suggests that people experience subjective fatigue when
mental resources are taxed (Cameron, 1973). Furthermore, self-
control resource depletion appears to coincide with subjectively
felt and physiological indicators of fatigue. Decrements in subjec-
tive fatigue have been shown in people engaging in depleting tasks
in ego-depletion experiments (e.g., Finkel et al., 2006; Friese,
Hofmann, & Wanke, 2008; Stewart, Wright, Hui, & Simmons,
2009). In addition, subjective fatigue evoked by engaging in cog-
nitively demanding tasks leads to elevated physiological indicators
of generalized fatigue and reduced performance on subsequent
tasks (Segerstrom & Nes, 2007; Wright et al., 2007, 2008). Such
effects suggest that fatigue may not be a mere indicator of ego
depletion but a mediator of the effects of self-regulatory resource
depletion on subsequent task performance (Muraven et al., 1998).
For example, the exertion of self-control requires effort, which can
lead to fatigue and, in turn, decreased capacity to exert self-control
in the future. Fatigue may also motivate people to conserve their
self-control resources when depleted (Muraven, Shmueli, & Burk-
ley, 2006). However, few studies have examined subjective fatigue
as a mediator of ego depletion. We aimed to examine the average
effect of ego depletion on fatigue as well as other subjective
indices of self-regulatory demand, such as effort and perceived
difficulty, in the present meta-analysis.
Decreased motivation may provide a further alternative expla-
nation for performance decrements on the second self-control task
in dual-task paradigm experiments. One proposed model is that the
ego-depletion effect is the result of reduced motivation to attain
task goals. This may occur if a person perceives insufficient
incentive to pursue the task goal, such as little reward or a lack of
value attached to the outcome, relative to the effort demand of the
task. The perceived imbalance between incentives and required
effort is likely to lead to a drop in motivation. People will persist
with tasks only so long as the reward or outcome is deemed worth
the effort; when it is not, they will lose motivation and tend to
desist. A motivation-only account of the ego-depletion effect
would therefore predict that decreased regulatory capacity was
independent of self-control resources. Evidence to support this
explanation is evident in the mental fatigue literature, in which
people are equally effective in performing short-term tasks, re-
gardless of their state of mental fatigue, provided incentives are
sufficient and their motivational state is high (Boksem, Meijman,
& Lorist, 2006; Lorist, Boksem, & Ridderinkhof, 2005; Tops,
Lorist, Wijers, & Meijman, 2004).
According to strength model theorists, a motivational account
for regulatory failure on tasks is not inconsistent with model
predictions. Proponents suggest that the effects of self-control
resource availability and motivation on task performance may be
interactive (Muraven & Baumeister, 2000). Reduced self-control
resources as a result of engaging in self-control tasks may lead
individuals to view task goals as unimportant due to the relatively
costly demand that doing the task will place on remaining re-
sources. This will result in reduced motivation on future tasks
requiring self-control. Consistent with this premise, people may be
able to overcome the debilitating effects of self-control resource
depletion if they are given sufficient incentive to do so. Research
has supported the moderating effect of motivational incentives,
such as rewards and increasing the importance of task outcomes,
on ego depletion (Muraven & Slessareva, 2003; Stewart et al.,
2009). Even though resources may be lowered through prior ex-
ertion of self-control, they may be only partially depleted, leaving
the potential for additional resources to be available (Muraven &
Baumeister, 2000). However, motivation may only temporarily
stem self-regulatory failure by causing people to tap into additional
self-regulatory reserves, and “after a certain point, fatigue becomes
insurmountable” (Baumeister et al., 2007, p. 353). A limited re-
source account of ego depletion therefore suggests that increased
motivation can only stave off self-control failure to the extent that
self-control resources remain available.
In summary, two competing explanations exist. A motivation-
only account suggests that impaired performance on self-control
tasks is purely a function of motivation and is not due to a limited
resource. According to this approach, fatigue induced by initial
self-control tasks leads to reduced motivation to perform subse-
quent tasks, probably due to the aversive nature of the tasks, but
when motivation is high, performance can be maintained. In con-
trast, the strength model proposes that engaging in initial self-
control tasks depletes self-control resources, at least partially,
leading to fewer resources being available to perform subsequent
tasks. However, increasing motivation to achieve a task goal may
lead people to commit more of their increasingly limited self-
control resources to engage in subsequent tasks, minimizing the
impairment of performance. In the present analysis, we aimed to
synthesize research that has examined motivation as a moderator
of the ego-depletion effect and establish whether motivational
manipulations, such as incentives, consistently minimize self-
regulatory failure brought about by initial depletion. In particular,
we make comparisons between depleted groups that receive mo-
tivational incentives and depleted groups that receive no incen-
tives. This will provide some evidence that motivation can help
overcome an ego-depleted state but may not provide sufficient
evidence to resolve the competing motivation-only explanation
and the motivation-plus-limited-resource explanation provided by
the strength model.
The self-regulatory failure observed in ego-depletion studies
could be the result of reduced self-efficacy. Although people may
view the goal of the task as attractive or important and believe they
could achieve the goal if they exerted the required effort, they may
perceive a reduced ability to reach it when in a depleted state.
There is limited research examining the role of self-efficacy in
ego-depletion studies. Wallace and Baumeister (2002) used bogus
competence feedback on the initial task in the dual-task paradigm
to induce high or low levels of self-efficacy. Identical patterns of
ego depletion were found for participants across feedback condi-
tions, leading Wallace and Baumeister to conclude that perceptions
relating to ability are unlikely to be implicated in self-control
resource depletion. This conclusion has been corroborated in stud-
ies that have found no relationship between self-reported self-
efficacy and ego depletion in dual-task paradigm experiments
(Baumeister et al., 2006; Finkel et al., 2006; Gailliot & Baumeis-
ter, 2007b). A possible reason for this is that reduced self-efficacy
in one sphere may not necessarily transfer to reduced perceptions
of ability toward a task in another, as in the dual-task paradigm.
Self-efficacy may therefore be inadequate as an explanation for
depletion because it does not transfer across task domains. Studies
that have included measures of self-efficacy concurrent with ego
depletion are included in the present meta-analysis to provide a
cumulative test of the effect of ego depletion on self-efficacy
The active regulation of emotion or mood has been shown to
deplete self-control resources and is a common means to invoke
ego depletion in the dual-task paradigm (Baumeister et al., 1998;
Bruyneel, Dewitte, Franses, & Dekimpe, 2009). This is because
regulating affect requires an individual to overcome the innate
tendency to display emotions in response to environmental stimuli.
However, negative affect may also be implicated in the develop-
ment of ego depletion. Tasks that require self-control are demand-
ing and frustrating and may induce a negative affective state (Leith
& Baumeister, 1996; Tice, Bratslavsky, & Baumeister, 2001). This
may compel a person to actively cope with or attempt to repair the
negative affect (Folkman & Moskowitz, 2000; Mayer & Gaschke,
1988). The coping process may reduce effort and motivation on
subsequent self-control tasks and thus lead to impaired task per-
formance. Studies including measures of affect as an additional
dependent variable in the dual-task paradigm have not generally
found a relationship between ego depletion and negative or posi-
tive affect (e.g., Baumeister et al., 1998; Bruyneel et al., 2009;
Muraven et al., 1998). There are exceptions; for example Ciarocco,
Sommer, and Baumeister (2001) and Stewart et al. (2009) found
significant postdepletion increases in negative affect after control-
ling for baseline relative to nondepleted controls. The authors of
these studies suggested that these increases served to indicate the
aversive nature of depleting tasks. In the present meta-analysis, we
provide a test of the effect of ego depletion on both negative and
positive affect. The analysis may help resolve any inconsistency in
the relationship between ego depletion and negative affect. A
strength model interpretation would predict a null effect, as ego
depletion is conceptualized as solely due to the depletion of
self-control resources. In contrast, a coping hypothesis would
predict a significant effect for ego depletion on negative affect.
Finally, self-regulatory failure is not expected to undermine or
alter levels of positive affect.
Experimenter Demand
Ego depletion may be an artifact of experimenter demand.
Poorer performance on the second self-control task in dual-task
paradigm experiments could be due to participants believing that
they have sufficiently fulfilled the experimenter’s demands after
completing the initial task. In order to rule out this alternative
explanation, studies have presented the tasks as separate experi-
ments (e.g., Baumeister et al., 1998; Burkley, 2008; Fischer, Gre-
itemeyer, & Frey, 2008; Seeley & Gardner, 2003; Vohs, Baumeis-
ter, & Ciarocco, 2005; Vohs & Faber, 2007) or unrelated tasks
(e.g., Johns, Inzlicht, & Schmader, 2008; H. M. Wallace &
Baumeister, 2002). Studies have also been conducted in which
each task was administered by a different experimenter (e.g.,
Richeson & Shelton, 2003; Richeson, Trawalter, & Shelton, 2005;
Vohs et al., 2008). The ego-depletion effect was found to be
consistent in these studies and serves to falsify the experimenter
demand explanation. We aimed to provide further confirmatory
support for these findings in the present analysis by including the
presentation of depleting tasks either as single or separate exper-
iments or by the same or different experimenters as moderators of
the ego-depletion effect.
Moderators of Ego Depletion
Although studies adopting the dual-task paradigm have gener-
ally supported the ego-depletion effect, the features of the tasks
that deplete self-control resources or serve to measure depletion
have not been systematically evaluated. Furthermore, the extent to
which second self-control task performance is a function of the
control condition used in the initial task is unknown. The features
of these tasks may lead to variations in the strength of the ego-
depletion effect and help resolve the inconsistencies observed in
some studies (Stillman et al., 2009; Wright et al., 2007, 2008).
Extraneous factors other than task features may also influence the
size of the ego-depletion effect. For example, individuals with
higher trait self-control may have an extended pool of self-control
resources to draw from, which would increase their self-control
capacity. We address these factors next.
Spheres of Self-Control
The dual-task paradigm permits an empirical test of the gener-
ality of the ego-depletion effect across different domains or
spheres (Baumeister & Vohs, 2007). Baumeister et al. (2007)
proposed that tasks requiring self-control be categorized into sev-
eral domains or spheres: (a) controlling attention, (b) controlling
emotions, (c) controlling impulses, (d) controlling thoughts, (e)
cognitive processing, (f) choice and volition, and (g) social pro-
cessing. Alternatively, tasks could be subsumed by more general
categories that reflect the global processes demanded by the task,
such as whether they require cognitive or affective processing.
Regardless of the classification system, levels of ego depletion are
expected to be equivalent in all spheres. This is important for the
strength model, as it would confirm the hypothesis that acts of
self-control draw energy from a common, global resource and that
self-control failure is domain general and not an artifact of a
particular sphere of task.
Notwithstanding the observed consistency of the ego-depletion
effect across tasks in these spheres of self-control (Baumeister et
al., 2007), there appears to be some inconsistency in the literature
as to what exactly constitutes a self-control task. A typical feature
of those tasks frequently adopted as depleting or dependent tasks in
the dual-task paradigm is the requirement for the effortful suppression
of an impulse or overriding of a habitual or dominant response.
However, questions remain as to whether difficult tasks involving
complex and challenging calculations (e.g., math problems and
analytical reasoning tasks) demand self-control resources.
Baumeister and colleagues (Baumeister et al., 1998; Muraven et
al., 1998) contended that such tasks do not necessarily deplete
self-control resources, because they involve the application of
well-learned algorithms or heuristics rather than the exertion of
overt control over the self to resist an impulse or override a
dominant response. As a consequence, these tasks have been used
as the nondepleting initial task in the dual-task paradigm (e.g.,
Muraven, Collins, & Nienhaus, 2002; Muraven, Shmueli, & Burk-
ley, 2006, Study 1; Muraven et al., 1998, Study 3).
However, studies have also adopted difficult and complex tasks
as the depleting task or dependent variable in dual-task paradigm
studies (e.g., Johns et al., 2008; Park, Glaser, & Knowles, 2008;
Schmeichel, 2007; Wright et al., 2008). Such tasks may demand
self-control resources for two reasons. First, they are often rated as
high in difficulty, effort, and unpleasantness, such that self-control
is required to resist the temptation to quit. For example, Wright et
al. (2008) demonstrated impaired performance on regulatory tasks
(e.g., incongruent Stroop color-naming task) as well as “nonregu-
latory” tasks (e.g., multiplication math task) after working on an
initial depleting task. There is also evidence in other literatures,
such as those for “mental” or “cognitive” fatigue (Ackerman &
Kanter, 2009) and vigilance (See, Howe, Warm, & Dember, 1995).
In these fields depletion effects have been demonstrated with tasks
that are high in difficulty but do not share the typical features of
self-control tasks. Second, tasks that require executive function
place demands on cognitive systems (e.g., the need to maintain and
update working memory). Such processes may share some features
of the tasks, such as response inhibition, typically used in ego-
depletion experiments. For example, memory updating tasks re-
quire people to constantly override the tendency to memorize
items in a particular order and apply a different rule. This is the
case in the reverse span memory tasks adopted by Schmeichel
(2007, Study 2) that required the memorization of digits in a
reverse order. This was considered more demanding of self-control
resources relative to a control condition task that required one to
merely hold information in memory. The evidence suggests that
tasks high in difficulty and complexity also serve to deplete
self-control resources.
These findings have two implications for the strength model.
First, it seems that effect of self-control depletion on task perfor-
mance may apply to a broader set of tasks and processes than
originally proposed, including difficult or complex tasks. This is
not inconsistent with the strength model, as such tasks require
regulatory effort to resist the temptation to quit or apply complex
rules that demand the inhibition of one process in favor of another.
Second, tasks vary in difficulty and complexity and, therefore, are
likely to vary in the extent to which they require self-control
resources (Muraven et al., 2002). An aim in the present review was
to examine whether this distinction acted as a moderator. That is,
we aimed to classify tasks used in the depletion of self-control
resources and as the dependent measure of self-control capacity in
studies adopting the dual-task paradigm according to the demand
they present to self-control resources (e.g., complexity) and to test
this classification as a moderator of the ego-depletion effect.
Control Condition
Experiments adopting the dual-task paradigm have typically
used a modified version of the depleting task that ostensibly does
not require self-control resources as the control condition. For
example, impulse control tasks require a person to override a
well-learned or spontaneous response and consciously apply effort
to perform the goal of the task. This might mean crossing out all
instances of a particular letter in a passage of text but only when
it is adjacent to certain other letters; resisting a tasty, tempting food
and eating a bland, less appetizing food instead; or inhibiting the
natural tendency to read the word rather than name the font color
in an incongruent Stroop color-naming task. In each case the
control condition simply requires that participants engage in an
easier, less effortful version of the depleting task in which partic-
ipants enact the dominant, impulsive response. So control partic-
ipants in the aforementioned examples would be asked to cross out
all instances of the letter without the inhibiting rules, given license
to taste tempting foods, and provided with a version of the Stroop
task in which the word and font color are congruent. Although
some of the “easier” versions of these tasks may require a modi-
cum of self-control to resist the urge to quit, as they are tedious and
boring, they are considered far less demanding and are not ex-
pected to tax self-control resources to the same degree as the
depleting versions.
However, some ego-depletion studies have adopted control
tasks other than easier versions of the depleting task. For example,
control participants have engaged in a task other than that admin-
istered to the depletion group, which is low in effort and ostensibly
does not tax self-control resources (e.g., Burkley, 2008, Study 3;
Muraven et al., 1998, Study 3). Alternatively, investigators have
required all participants to engage in the same initial task (e.g.,
describing a target person from a minority group), which would
require self-control only for people possessing a specific individ-
ual difference variable (e.g., motivation to avoid prejudice; e.g.,
Gailliot, Plant, et al., 2007; Gordijn, Hindriks, Koomen, Dijkster-
huis, & Van Knippenberg, 2004; Muraven, 2008b; Park et al.,
2008). There are also studies in which control participants did not
engage in an initial task at all but sat passively before engaging in
the dependent task (e.g., Baumeister et al., 1998, Study 3; Burkley,
2008, Study 2). Given the variety of approaches to evoking ego
depletion and the tasks adopted in the control condition, it is
important to establish whether the ego-depletion effect varies
according to the type of task to which the depleting task is being
compared. In the present meta-analysis, the type of control group
is used as a moderator of the ego-depletion effect to test this
Personality and Individual Differences
The strength model focuses on state depletion of self-control
resources. Self-control strength is viewed as vulnerable to deple-
tion after acute bouts of self-regulatory effort; this subsequently
limits short-term self-control capacity. However, numerous
capacity-based theories of self-control also conceptualize self-
control as a dispositional, traitlike construct that differs across
individuals (Funder, Block, & Block, 1983; Metcalfe & Mischel,
1999; Muraven & Baumeister, 2000; Schouwenburg, 2004; Tang-
ney et al., 2004; Wills & Dishion, 2004). For example, delay of
gratification is viewed as a generalizable capacity to forgo short-
term, alluring rewards in favor of greater, long-term rewards
(Mischel, 1996; Mischel et al., 1989). This view has been incor-
porated into the strength model. Although all individuals are
vulnerable to state depletion of self-control resources, individuals
are proposed to differ in their overall self-control capacity
(Baumeister et al., 2006). This implies that people high in dispo-
sitional self-control will have more resources at their disposal.
Such individuals will have more resources remaining after engag-
ing in a self-control task of a given duration and resource demand
than will individuals lower in trait self-control. Trait self-control
may therefore serve to insulate a person from the depleting effects
of self-control tasks and moderate the ego-depletion effect. The
proposed interaction between dispositional self-control measured
on psychometric instruments and ego depletion has been tested
empirically, and results are inconclusive. Some studies have found
a clear interaction effect (Dvorak & Simons, 2009; Gailliot &
Baumeister, 2007b; Gailliot, Schmeichel, & Maner, 2007), and
others have found no interaction (Gailliot & Baumeister, 2007b;
Stillman et al., 2009). Although a resolution to these inconsistent
findings through meta-analytic synthesis is needed, there are only
a handful of effect sizes and few studies report the ego-depletion
effect in both high and low dispositional self-control groups.
In addition, few studies have investigated the moderating effect
of other traits and individual difference variables on ego depletion.
Ego depletion has been shown to be positively associated with
high other-orientations and low self-monitoring (Seeley & Gard-
ner, 2003; Wan & Sternthal, 2008), higher levels of fluid intelli-
gence (Shamosh & Gray, 2007), and higher levels of consideration
of future consequences–immediate (Joireman, Balliet, Sprott,
Spangenberg, & Schultz, 2008). The mechanisms behind these
individual difference moderators can be explained through greater
motivation to allocate self-regulatory resources among those with
higher levels of the trait. For example, superior performance on
self-control tasks under depleting conditions for individuals high
in other orientation and low in self-monitoring has been attributed
to greater motivation to meet the expectations of others and com-
ply with normative standards (Seeley & Gardner, 2003; Wan &
Sternthal, 2008). Individuals possessing fluid intelligence and con-
sideration of future consequences–immediate may be predisposed
to be more susceptible to the situational demands of tasks leading
them to consume more self-control resources in the first task in the
dual-task paradigm than are individuals in which these traits are
absent (Joireman et al., 2008; Shamosh & Gray, 2007). In terms of
mechanisms, individuals high in these traits may be more moti-
vated to succeed and more likely to invest effort in the initial task
at the expense of making resources available for subsequent self-
control efforts. These findings indicate that the examination of
individual differences in self-control may shed light on the factors
that magnify or diminish the ego-depletion effect and provide
information on underlying processes and boundary conditions. The
inclusion of such moderators in future investigations using the
dual-task paradigm is warranted.
Extending the Model
The view that self-control is akin to a limited resource in the
strength model has given rise to additional related hypotheses. It is
proposed that people will tend to conserve their resources when
these resources are scarce, are able to improve their self-control
capacity through practice or training, and are able to recover their
self-control resources through rest or supplementation with glu-
cose. These hypotheses are presented in the next section.
The ego-depletion effect may be the result of a complete deple-
tion of finite self-control resources; such depletion negates any
possibility of subsequent acts of self-control. An alternative hy-
pothesis is that self-control tasks only partially deplete resources
and ego depletion occurs because people are either unable or
unwilling to draw further from their reserves (Baumeister & Vohs,
2007; Muraven & Baumeister, 2000). One reason for this is that
people conserve their limited self-control resources, particularly
when they expect future exertion (Baumeister & Heatherton, 1996;
Baumeister, Muraven, & Tice, 2000; Muraven & Baumeister,
2000). This is an adaptive strategy from a resource allocation point
of view, as people are motivated to retain a residual level of their
self-control resources in anticipation of future need. According to
the strength model analogy, this is akin to an athlete conserving
energy for a final effort toward the end of a race. Experiments in
which participants were informed that they would be required to
perform an additional third self-control task after the second self-
control task have supported this hypothesis (Muraven, Shmueli, &
Burkley, 2006; Tyler & Burns, 2009). Ego-depleted participants
expecting another self-control task showed impaired performance
on subsequent self-control tasks relative to depleted controls with
no such expectations.
The conservation hypothesis is also consistent with a motiva-
tional account of ego depletion. Low self-control resources make
the prospect of engaging in future self-control tasks more daunting,
as it is more costly to allocate resources when they are scarce than
when they are plentiful. As a consequence, motivation to engage in
future self-control tasks will be decreased due to the high cost
involved and low importance of the task goal relative to the
importance of conserving self-control resources. This tendency to
conserve is exacerbated when the prospective demands of future
self-control are very high, such as when the prospect of a third
self-control task is highlighted in the dual-task paradigm. The
tendency to conserve will be greater and motivation toward the
task will be diminished due to the expected future load. In this
article, we aim to test whether the conservation hypothesis is
supported across ego-depletion studies. We meta-analyze ego-
depletion studies employing the dual-task paradigm that have
included conditions in which participants were told to anticipate a
third self-control task.
In the previous section we introduced the hypothesis that train-
ing on self-control tasks improves self-control capacity and atten-
uates the ego-depletion effect (Baumeister et al., 1998). According
to the strength model, just as a muscle increases in strength with
training, the capacity to exert self-control will be heightened after
repeated practice on self-control tasks. Support for this hypothesis
has been provided in studies using self-control tasks in different
spheres to train self-control. For example, participants required to
engage in everyday tasks requiring self-control (e.g., improving
posture, regulating mood, monitoring eating habits, and avoiding
colloquial language) for a period of 2 weeks performed signifi-
cantly better than untrained controls in a subsequent dual-task
paradigm (Gailliot, Plant, et al., 2007; Muraven et al., 1999). Oaten
and Cheng (2006a, 2006b, 2007) corroborated these results in a
series of studies in which participants engaged in formal programs
of academic study, physical exercise, and financial monitoring
over a period of months. Participants engaging in the programs
also reported being more effective in managing other everyday
behaviors requiring self-control (Oaten & Cheng, 2006a, 2006b).
These findings demonstrate that regular practice can help attenuate
the ego-depletion effect. The mechanism for these improvements
may be through increases in the amount of self-control resources
available, an “extended pool” explanation, or, alternatively,
through improved efficiency on the task, perhaps in the develop-
ment of more effective self-control “skills.” The present analysis
provides a cumulative test of the training hypothesis across studies.
Periods of rest or relaxation may help to restore self-control
resources after depletion and minimize the deleterious effects of
depletion on subsequent task performance (Baumeister & Heath-
erton, 1996; Muraven & Baumeister, 2000). This is consistent with
the muscle metaphor of the strength model: Muscles require a
period of recovery after exertion before they can apply further
force. Studies have tested this hypothesis by introducing rest or
relaxation periods between tasks in the dual-task paradigm (Oaten,
Williams, Jones, & Zadro, 2008; Tyler & Burns, 2008). Consistent
with the hypothesis, findings indicated that ego-depleted individ-
uals given the opportunity to rest or relax exhibited superior
second-task performance relative to nonrested depleted controls.
Of note, Tyler and Burns (2008) found evidence for a “dose
effect,” such that the restoration of self-control capacity is propor-
tional to the duration of the recovery period. Given this finding, it
follows that variability in the interim period between tasks in the
dual-task paradigm may account for some of the variability in the
ego-depletion effect across studies. For example, experiments in
which participants are required to complete questionnaires or take
a break during the intertask interim period may provide greater
opportunity to recover self-control capacity relative to experiments
in which participants proceed immediately to the second task. In
this article, we test the recovery hypothesis by subjecting studies
including rest or relaxation periods in a dual-task paradigm to a
meta-analytic synthesis. As an additional test, we examine whether
the inclusion of an intertask interim period moderates the ego-
depletion effect.
Glucose and Glucose Supplementation
In search for physiological mechanisms for self-control resource
depletion, Gailliot and coworkers (Gailliot & Baumeister, 2007a;
Gailliot, Baumeister, et al., 2007) provided preliminary evidence
that blood glucose may be the control mechanism for the depletion
of self-control reserves. These studies demonstrated that ego de-
pletion coincides with decreases in blood glucose and glucose
supplementation attenuates the ego-depletion effect relative to a
sweetened placebo (DeWall, Baumeister, Gailliot, & Maner, 2008;
Dvorak & Simons, 2009; Gailliot, Baumeister, et al., 2007; Gail-
liot, Peruche, Plant, & Baumeister, 2009; Masicampo & Baumeis-
ter, 2008). On the basis of these findings, Gailliot and coworkers
suggested that the strength model of self-control was more than
just a metaphor and that self-control resource depletion occurred
concurrent with the utilization of fuel substrates in the body. In the
present meta-analysis, we test the consistency of blood glucose as
an analogue for self-control depletion and the potential of glucose
supplementation to moderate the ego-depletion effect.
The Present Analysis
The past decade has seen a proliferation in research on self-
regulatory failure from the perspective of the strength model, and
a cumulative synthesis of these research findings is timely. At the
most basic level, the analysis will be expected to summarize the
overall effect of self-control resource depletion on task perfor-
mance. A cursory glance at trends in the literature and narrative
reviews suggests that the effect will be present, notwithstanding a
minority of studies that have found nonsignificant effects. How-
ever, it is important to evaluate the extent of the variability in the
effect and identify the moderator variables that may resolve this
variability. The evaluation of these moderators may assist in de-
termining the adequacy of the strength model in explaining ego
depletion, resolving any inconsistencies in the literature, and eval-
uating whether competing or complementary explanations can
offer insight into the variation in the effect.
Overall Ego-Depletion Effect
The effect size of interest is the overall averaged effect of ego
depletion on self-control task performance corrected for sampling
error variability across published studies using the dual-task ex-
perimental paradigm. The effect of ego-depleting self-control tasks
on participants’ second task performance relative to that of non-
depleted controls will be the key dependent variable. The strength
model predicts that the average ego-depletion effect will be sig-
nificant and homogeneous across studies.
Additional Dependent Variables
We test the effect of ego depletion on seven additional depen-
dent variables: effort, positive affect, negative affect, perceived
difficulty, subjective fatigue, self-efficacy, and blood glucose. In
keeping with the highly aversive and effortful nature of self-
control tasks, significant effect sizes for the effort, perceived
difficulty, subjective fatigue, and negative affect variables are
expected. Consistent with previous tests of ego depletion on pos-
itive affect, null or weak effects for ego depletion on this variable
are expected. The effect of ego depletion on self-efficacy is less
consistent and has not been frequently tested. A plausible hypoth-
esis is that self-efficacy may be implicated in the ego-depletion
effect because reduced resources may lower estimates of future
ability to exert self-control, although this has not been supported
empirically. Finally, the analysis tests whether self-regulatory fail-
ure brought about by ego depletion coincides with reduced blood
glucose levels. This may provide evidence that substrate use serves
as a physiological indicator of self-control resource depletion.
Analysis of Moderators
Several moderators hypothesized to magnify or diminish the
overall ego-depletion effect are included in the analysis. In this
article, we test whether the sphere of depleting self-control task in
dual-task paradigm experiments moderates the ego-depletion ef-
fect. Baumeister et al. (2007) proposed that self-control tasks from
different spheres would be equally depleting because self-control
draws from a single, global resource. An alternative hypothesis is
that some spheres of self-control may be more demanding of
self-control resources than others, so the moderator analysis may
identify spheres that place a greater burden on self-control re-
sources. We also evaluate whether the sphere of the dependent task
used as an index of ego depletion moderates the effect. Although
little variance in the ego-depletion effect is expected across these
dependent tasks, it is possible that some of these tasks may place
fewer demands on self-control resources or be less effortful than
others, leading to less of a decrement in performance. We also test
the variability of the ego-depletion effect across frequently used
depleting and dependent self-control tasks. This information will
provide some indication as to whether these types of task are
equally effective in inducing and measuring ego depletion.
The hypothesis that tasks vary in the extent to which they deplete
self-control resources has not been studied systematically, and the
present study is the first to synthesize studies on self-control
resource depletion according to task domain and type. In addition,
we evaluate whether using depleting and dependent tasks from the
same or different spheres in the dual-task paradigm moderates the
ego-depletion effect. If the proposed generality of the ego-
depletion effect holds, the size of the effect should be invariant
regardless of whether or not the depleting and dependent tasks are
“matched” on task sphere. This analysis also provides an indica-
tion of the extent to which these studies provide an effective test of
the generality hypothesis.
Task complexity is also included as a moderator. The aim in this
analysis is to resolve the apparent contradiction in the literature as
to whether tasks that are difficult or challenging, but do not
ostensibly require overriding a well-learned response, deplete self-
control resources. In addition, it is expected that the duration of
depleting task in the dual-task paradigm will be linearly related to
the size of the ego-depletion effect. We also test whether present-
ing tasks as single or separate experiments or by the same or
different experimenters moderates the overall ego-depletion effect.
The purpose in these analyses is to rule out the alternative expla-
nation that the ego-depletion effect is due to experimenter demand.
The analysis evaluates whether the dependent task is a behav-
ioral or self-report measure of ego depletion. We expect the
ego-depletion effect to be significant in both groups, but a logical
prediction would be that the ego-depletion effect will be inflated in
the group of studies using self-reported dependent variables due to
the increased measurement error associated with such measures
(Westholm, 1987).
The effect of the nature of the task used in the control condition
alongside the depleting task is also be evaluated as a moderator.
This analysis provides evidence on whether use of a modified,
“easier” version of the self-control task, a different task altogether,
or the same task alongside an individual difference variable that
renders the task more taxing for some individuals is equally
effective as use of comparison groups in ego-depletion experi-
ments. We also test whether cognitive processing tasks that require
complex processing result in greater ego depletion than tasks that
are simpler and present fewer demands. Finally, we compare the
ego-depletion effect in tests originating in the dominant Baumeis-
ter laboratory relative to tests from other laboratories. No differ-
ence is expected in the overall ego-depletion effect across the
laboratory moderator groups.
Testing Strength Model Hypotheses
In this article, we conduct a meta-analysis of studies that have
tested additional hypotheses derived from the strength model: the
conservation, training, and recovery hypotheses. For the conser-
vation hypothesis, we test whether the expectation of future de-
mands on self-control resources will diminish self-regulatory ca-
pacity indicating the tendency to conserve resources. The role of
using incentives and highlighting the importance of task outcomes
as means to reduce the ego-depletion effect is also evaluated. This
evaluation determines whether motivation mitigates the self-
regulatory failure brought about by engaging in initial self-control
tasks. For the training hypothesis, we meta-analyze studies that
have tested the ego-depletion effect in participants provided with a
period of training on self-control tasks prior to engaging in the
dual-task paradigm. For the recovery hypothesis, we meta-analyze
studies that have included a period of rest or relaxation or an
interim period between the first and second self-control tasks in the
dual-task paradigm. We predicted that the ego-depletion effect
would be attenuated through the provision of motivation-
enhancing strategies, training on self-control tasks, or a recovery
period after the initial depleting task. By analogy, the ego-
depletion effect was expected to be exacerbated if there was an
expectation of future acts of self-control. Finally, the role of
glucose supplementation as a means to mitigate the deleterious
effects of self-control resource depletion on subsequent task per-
formance is examined. It was anticipated that ego-depleted partic-
ipants’ self-control task performance would be improved by the
provision of glucose relative to a sweet placebo.
Literature Search
Published research articles were located via a search of elec-
tronic databases—ERIC, Embase, ISI Web of Science (Science
Citation Index Expanded, Social Science Citation Index Ex-
panded), Medline, PsycARTICLES, and PsycINFO—covering the
literature from 1998 (the year of Baumeister et al.’s and Muraven
et al.’s initial research articles on ego depletion) until April 1,
2009. The following search terms in various combinations were
used for all searches: ego depletion, ego energy, self-control, and
self-regulation, with subterms depletion, failure, limit, resources,
and strength. The reference sections of the retrieved articles con-
sidered for this review were scrutinized for additional studies. We
also examined the reference sections of key narrative reviews of
the literature on ego depletion and the strength model of self-
control (Baumeister, 2002, 2003; Baumeister et al., 2000, 2006;
Baumeister, Sparks, Stillman, & Vohs, 2008; Baumeister & Vohs,
2007; Baumeister et al., 2007; Gailliot & Baumeister, 2007a;
Muraven & Baumeister, 2000; Schmeichel & Baumeister, 2004;
Vohs & Baumeister, 2004). In addition, we searched for articles in
key social psychology journals and their repository of articles
published online in advance of print.
Inclusion Criteria
Studies were required to provide an experimental test of the
ego-depletion effect using variants of the dual-task paradigm out-
lined by Baumeister et al. (1998). For the overall ego-depletion
effect, studies had to include a quantifiable measure of task per-
formance for the second self-control task in the experimental
group that received the ego-depleting task and a nondepleted
control group. Studies that included a continuous or discrete (e.g.,
median split) individual difference moderator of the ego-depletion
effect were considered eligible if effect size data for the depleted
and nondepleted groups were available for the main effect of ego
depletion independent of the moderator (e.g., DeWall, Baumeister,
Stillman, & Gailliot, 2007, Study 4; Joireman et al., 2008, Study 3;
Richeson et al., 2005, Study 1; Seeley & Gardner, 2003). In some
cases, the initial depleting task was identical for all participants
and depletion was evoked on the basis of an individual difference
variable. For example, Gailliot, Plant, et al. (2007) divided their
sample into those with high or low scores on internal motivation to
respond without prejudice and then presented all participants with
a task requiring them to describe a homosexual target and avoid
stereotypical statements. The task was expected to deplete self-
control resources only in low-motivation participants, because
suppressing stereotypes was considered difficult and effortful for
people who did not regularly attempt to control prejudiced
thoughts. In such cases the individual difference variable served as
the depletion condition (e.g., Gailliot, Plant, et al., 2007; Seger-
strom & Nes, 2007; Vohs et al., 2005). To ensure that the use of
individual difference variables to deplete self-control resources did
not bias the overall ego-depletion effect, we included this as an
additional moderator variable in the meta-analysis. Some studies
included experimental manipulations expected to moderate the
ego-depletion effect (e.g., Tyler & Burns, 2008; H. M. Wallace &
Baumeister, 2002; Wan & Sternthal, 2008). In such cases, the
ego-depletion effect was calculated for depleted and nondepleted
participants in the condition where the moderator was absent. This
represents a simple, unattenuated test of the ego-depletion effect.
For example, Tyler and Burns (2008) introduced an experimental
condition of relaxing music between the initial and second tasks in
the dual-task paradigm to aid relaxation; control participants re-
ceived no music. In this case, the ego-depletion effect was calcu-
lated from the dependent task performance of depleted and non-
depleted individuals assigned to the music-absent condition.
Nonexperimental studies assessing the effect of self-reported,
trait measures of self-control on task performance and studies that
did not use the dual-task paradigm or include a nondepleted
control group were rejected (e.g., Fennis, Janssen, & Vohs, 2009,
Study 6; Finkel & Campbell, 2001; Muraven, 2008a; Muraven,
Collins, Shiffman, & Paty, 2005; Muraven et al., 1998, Study 4;
Neubach & Schmidt, 2008; Oaten et al., 2008; Schmeichel & Zell,
2007; Schmidt, Neubach, & Heuer, 2007; Tangney et al., 2004;
J. C. Wallace, Edwards, Shull, & Finch, 2009). The only two
exceptions to this were studies that included nonbehavioral or
judgment measures in place of the second task and studies that did
not include a nondepletion control group but provided tests of the
conservation, training, and recovery hypotheses from the strength
model. Studies using a nonbehavioral or judgment dependent
variable were included because such measures not only serve as an
analog of ego depletion but also reflect decisions likely to place a
demand on self-control resources (e.g., Burkley, 2008; Fischer,
Greitemeyer, & Frey, 2007). However, such measures may be
unsuitable as the initial depleting task in the dual-task paradigm.
To check that the inclusion of these studies did not bias the overall
ego-depletion effect, we also included dependent measure type
(behavioral vs. nonbehavioral or judgment) as a moderator of the
overall ego-depletion effect.
Finally, studies had to contain sufficient statistical information,
such as cell means and standard deviations, Fratios, tstatistics,
zero-order correlations (r), or effect size statistics (e.g., Cohen’s d,
), to calculate an estimate of effect size. Missing data for studies
that were otherwise eligible were requested by contacting the
authors. Finally, eligible studies were systematically screened for
duplicates to eliminate bias due to duplicate study effects (Wood,
Meta-Analytic Strategy
We used Hunter and Schmidt’s (1994) methods for meta-
analysis to correct effect sizes for sampling error variability.
Monte Carlo simulation studies have supported the adoption of a
random effects model because it permits the generalization of
corrected effect sizes to the population (Field, 2003; Hunter &
Schmidt, 2000; Kisamore & Brannick, 2008). The effect size
metric employed in the current analysis was Cohen’s d, which
represents the standardized mean difference score for experimental
(ego depletion) and control (nondepletion) groups in studies adopt-
ing the dual-task paradigm. Effect sizes were calculated directly
from the means, standard deviations, and sample sizes for the
experimental and control groups wherever possible. Standard for-
mulas were used to compute the effect size statistic from Fratios,
ttests, or zero-order correlation coefficients (DeCoster, 2004;
Hullett & Levine, 2003). In cases where multiple methods for
effect size calculation were available, we used the test most closely
based on the means, standard deviations, and sample sizes
(DeCoster & Claypool, 2004).
In addition to producing the averaged overall effect size cor-
rected for sampling error (d
), we calculated 95% confidence
intervals (CI
) with the standard error of the mean effect size to
test the accuracy of effects. We also computed the fail-safe sample
size (N
), which represents the number of unpublished studies
with null findings that would have to exist in the researchers’ file
drawers to reduce the effect size to a zero value (Rosenberg, 2005).
If the number of “null finding” tests of an effect is sufficiently
large, the researcher can be confident that the chance of such a
number of studies existing is negligible. Rosenberg suggested a
critical value of 5N10 for the N
In addition, the percentage variance in the effect sizes across
studies attributed to within-study sampling error variability rela-
tive to overall between-study variance in the effect size was
calculated. This ratio of variance is an important first step in
establishing the homogeneity of the effect size, that is, whether the
vast majority of the variance in the effect across studies can be
accounted for by sampling error (Aguinis & Pierce, 1998). Hunter
and Schmidt (1994) have proposed that the proportion of variance
attributed to within-study sampling variance should exceed 75%
for an effect size to be considered homogeneous. A formal test of
the relative homogeneity of an effect is given by Cochran’s (1952)
Qstatistic. The Qstatistic is a relatively conservative test and can
lead to Type II errors (i.e., increased probability of accepting as
homogeneous sets of studies that have substantial heterogeneity),
so we adopted the 75% rule in the cases where the Qstatistic was
We also conducted our meta-analyses using a fixed-effects meta-
analytic method (Hedges & Olkin, 1985). Effect sizes and confidence
intervals computed for the overall ego-depletion effect and the moderator
group, additional variables, and strength model hypotheses analyses were
not appreciably different from those produced with a random effects
model. Data for the fixed-effects analyses are available on request from the
first author.
significant (Sagie & Koslowsky, 1993; Sa´nchez-Meca & Marı´n-
Martı´nez, 1997).
As the number of studies (k) varies across meta-analyses, the Q
statistic cannot be compared across analyses, so we also calculated
the I
statistic and its confidence interval as an alternative (Hig-
gins, Thompson, Deeks, & Altman, 2003). This represents a mea-
sure of true heterogeneity in the effect size in question expressed
as a percentage and is easily interpretable, with levels of 25%,
50%, and 75% representing low, medium, and high levels of
heterogeneity respectively (Higgins & Thompson, 2002). Should
the I
value exceed 25% with a wide confidence interval that does
not have a zero lower limit, it is likely that substantial heteroge-
neity in the effect size exists (Huedo-Medina, Sa´nchez-Meca,
Marı´n-Martı´nez, & Botella, 2006). This would suggest that the
effect may be influenced by extraneous moderator variables.
Finally, we treated the effect size data for the overall ego-
depletion effect and all moderator and additional variable suba-
nalyses for statistical outliers. We adopted Grubbs’ (1950) test to
identify outliers in accordance with Barnett and Lewis’s (1994)
recommendations. Identified outliers were set at the value of the
next closest effect size in the data set.
Moderator Coding
Sphere of self-control task. We conceptualized sphere of
depleting and dependent tasks in three different ways: as specific
and distinct spheres, as suggested by Baumeister et al. (2007); as
global spheres, according to the demands placed on cognitive or
affective processing systems;
and as tasks frequently used in the
dual-task paradigm. Baumeister et al.’s suggested spheres of self-
control are based on the features of the task: controlling attention,
emotions, thoughts, and impulses, cognitive processing, choice
and volition, and social processing. Tasks requiring attention con-
trol involved focusing attention and disregarding distractions, such
as watching the central figure in a video while ignoring words
displayed in a corner of the screen (e.g., Fischer et al., 2008, Study
1; Schmeichel & Vohs, 2009, Study 2). Controlling emotions tasks
demanded the active suppression of emotional responses, such as
requiring participants to avoid expressing emotions when watching
emotionally appealing or aversive videos (e.g., Baumeister et al.,
1998, Study 3; Hofmann, Rauch, & Gawronski, 2007). Impulse
control tasks demanded that participants resist gratifying courses
of action or override well-learned habits, such as resisting tempting
foods (e.g., DeWall et al., 2007, Study 1; Geeraert & Yzerbyt,
2007, Study 1b) or suppressing the urge to name the target word
instead of the typeface color in Stroop color-naming tasks (e.g.,
Bray, Ginis, Hicks, & Woodgate, 2008; Webb & Sheeran, 2003,
Study 1). Controlling thoughts tasks demanded that participants
suppress unwanted thoughts; an example is Wegner, Schneider,
Carter, and White’s (1987) paradigm that requires participants to
avoid thinking of a “white bear” (e.g., Burkley, 2008, Studies 3
and 4; Tyler, 2008, Study 4). Tasks requiring choice or volition
involved participants making a choice or decision between options
in numerous contexts (e.g., consumer choices or choosing to write
an essay in a forced-choice paradigm; see, e.g., Baumeister et al.,
1998, Study 2; Bruyneel, Dewitte, Vohs, & Warlop, 2006, Studies
1, 2, and 3). Cognitive processing tasks involved exerting effort to
maintain a high degree of executive functioning or processing
information, as when working on challenging anagram, memory
span, or counting tasks (e.g., Park et al., 2008; Schmeichel, 2007,
Study 3; Wright et al., 2007, Studies 1 and 2). Social processing
tasks required the processing of social information that might
involve searching for appropriate social cues, such as when sup-
pressing stereotypes, resisting persuasion, or engaging in high-
maintenance social interactions (e.g., Burkley, 2008, Study 1;
Finkel et al., 2006; Gordijn et al., 2004, Studies 2 and 4; Richeson
& Shelton, 2003). Moderator coding was initially conducted by the
three lead authors. An independent judge familiar with self-control
tasks and the dual-task paradigm then performed the classification
independently. The agreement of the judge’s classification with
those of the coauthors was high (␬⫽.81, p.001). Discrepancies
were discussed with the judge, and a consensus decision was made
to resolve the ambiguity.
Depleting and dependent tasks were also subjected to a global
classification according to the extent to which they placed de-
mands on affective or cognitive processing or a combination of the
two. Tasks requiring the control of emotions and impulses were
classified as affective. Controlling attention and thoughts, choice
and volition, and cognitive processing tasks were classified as
cognitive. Finally, social processing tasks were classified as re-
quiring both affective and cognitive processes. The latter was
justified on the basis that such tasks demand a degree of cognitive
control or regulation but also the suppression of impulses or
emotions. For example, tasks that require participants to identify
relational cues involve a cognitive process to focus attention on the
appropriate cue and an affective process to avoid the impulse or
temptation to fall back on heuristic processing (Tyler, 2008).
Frequently used tasks. We also created a moderator variable
for tasks frequently used to deplete and measure self-control in the
dual-task paradigm. Tasks that had a frequency of use greater than
10 in the sample of studies were considered eligible. Frequently
used depleting tasks were the video-watching affect regulation,
video-watching attention control, crossing out letters, modified
Stroop (1935), and Wegner et al.’s (1987) “white bear” thought-
control tasks. Frequently used dependent tasks included the hand-
grip, solvable anagram, food taste test, math or mental arithmetic,
and modified Stroop (1935) tasks.
A taxonomy of self-control-depleting and self-control-
dependent tasks was developed to summarize the coding of tasks
in each sphere of self-control. The taxonomy, provided in Appen-
dix A, is organized with depleting tasks in the top half and
dependent tasks in the bottom half.
The tasks are categorized
according to Baumeister et al.’s (2007) spheres of self-control at
the subordinate level, with global affective, cognitive, and com-
bined affective and cognitive spheres as higher order categories. A
brief description of the task, studies that have adopted the task, and
frequency of task use are included in each cell of the table. The
taxonomy is intended as a resource with which to identify the
category and frequency of use of the self-control tasks adopted as
depletion and dependent tasks in dual-task paradigm experiments.
Duration of depleting task. The longer the duration of the
depleting self-control task in the dual-task paradigm, the greater
the expected impairment of performance on the second task. This
We thank an anonymous reviewer for this suggestion.
Appendices A, B, and C are made available as online supplementary
materials (
is because the amount of self-control resource consumed is as-
sumed to be proportional to the time spent exerting self-control.
We extracted depleting task duration in minutes from studies that
reported these data. In cases where data on task duration were
missing but a standardized depleting task was used (e.g., modified
Stroop task), the task duration was inferred from the average
duration for that task across studies reporting duration data.
Interim period. As few studies reported the precise time
period between the two tasks in the dual-task paradigm, we aimed
to test whether the type of intertask interim activity moderated the
ego-depletion effect. Studies were therefore classified on the basis
of the activities performed during the interim period. Studies were
classified as those that reported no activities during the interim
period, those that required participants to complete questionnaires
(e.g., mood scales, manipulation check items), and those that
required participants to engage in a filler task or reported a rest
period between tasks. We predicted, in accordance with the recov-
ery hypothesis, that studies reporting no interim period would
provide less opportunity for participants to recover their self-
control resources, leading to a larger ego-depletion effect. In
contrast, participants required to complete questionnaires or filler
tasks or prescribed a period of rest between tasks were expected to
have greater chance for recovery and a comparatively smaller
ego-depletion effect.
Experiment presentation. We also tested whether presenting
tasks in the dual-task paradigm as single or separate experiments
or by the same or different experimenters moderated the overall
ego-depletion effect. Studies making an explicit statement that the
tasks were presented as separate experiments were coded as one
moderator group, and those making no explicit distinction were
assumed to have presented the tasks as a single experiment and
were coded as the other moderator group. In addition, studies that
reported using different experimenters to administer the self-
control tasks formed one moderator group, and studies using the
same experimenter formed the other moderator group.
Type of dependent task. We conducted a moderator analysis
to evaluate whether nonbehavioral or judgment dependent self-
control measures in the dual-task paradigm were as vulnerable as
behavioral measures to the ego-depletion effect. Studies were
therefore coded according to whether they used a behavioral
measure, such as performance of a self-control task, or used a
nonbehavioral or judgment measure that was indicative of self-
control effort. Examples of nonbehavioral or judgment dependent
measures include subjective ratings of a target person, with re-
sponses analyzed for prejudiced or aggressive responses (e.g.,
Muraven, 2008b; Stucke & Baumeister, 2006), or self-reported
evaluations that require suppression of well-learned tendencies,
such as being modest and self-effacing rather than being narcis-
sistic or exaggerating ability (e.g., Fischer et al., 2007; Vohs et al.,
2005, Study 8).
Matched depleting and dependent tasks. Studies were
coded according to whether the depleting and dependent self-
control tasks in the dual-task paradigm were from the same or
different spheres in accordance with to the previously cited criteria
for specific and global spheres of self-control. We therefore de-
veloped two moderator groups. Studies that adopted depleting and
dependent tasks from the same specific sphere (controlling emo-
tions, thoughts, impulses, and attention, choice and volition, cog-
nitive processing, and social processing) or global sphere (affec-
tive, cognitive, or combined affective and cognitive) were coded as
matched, and those adopting tasks from different spheres were
coded as unmatched.
Control condition. We coded studies according to the type of
control condition that was adopted in the dual-task paradigm. The
majority of studies used modified, easier versions of the depleting
task that did not require the overriding of an impulsive or well-
learned response. Some studies used an alternative task that was
ostensibly easier and required less self-control than did the deplet-
ing task. For example, Muraven et al. (1998, Study 3) used Wegner
et al.’s (1987) white bear thought control paradigm to deplete
self-control resources and compared these participants with a con-
trol group that solved math problems. Muraven et al. argued that
the active suppression of unwanted thoughts demanded more self-
control resources than math problems, as the latter did not require
the active suppression of a dominant response. It is, nevertheless,
possible that the math problems still required some degree self-
control in that participants had to resist the urge to quit. The
adoption of such tasks in the control condition may require a
degree of self-control but less than that required by the depleting
tasks in the experimental condition. This may serve to moderate
the ego-depletion effect, so we coded studies into moderator cat-
egories according to their use of easier versions and alternative
tasks in the control condition. We also coded studies that used the
same initial task but included an individual difference variable to
evoke depletion. For example, Gordijn et al. (2004) segregated
their sample into individuals scoring low and high on a motivation
to respond without prejudice measure, and participants in both
groups were then asked to describe a person from a minority group
while suppressing stereotypes. Consistent with predictions, the
low-motivation group exhibited greater ego depletion on the sec-
ond task than did the high-motivation group. Some studies adopted
a design in which the initial task was absent for control participants
(Baumeister et al., 1998, Study 3; Burkley, 2008, Study 2). How-
ever, too few studies (k2) were available for us to include this
as a moderator category.
Task complexity. Although self-control tasks share a similar
domain or sphere, they are quite variable in their specific features,
making it difficult to evaluate whether these features may be more
demanding than others of self-control resources. For example, it is
difficult to establish whether resisting tempting foods in an osten-
sible taste test is more demanding or effortful than resisting the
discomfort in the forearm muscles while holding a handgrip ap-
paratus. However, some tasks in the cognitive processing domain
with clear common features have been classified as more or less
demanding due to their complexity and the type of processing
required. Schmeichel’s (2007) experiments on the resource-
depleting effect of tasks requiring executive functioning identified
tasks that varied in the extent and number of processes required; it
was proposed that the level of complexity was proportional to the
degree of ego depletion evoked. For example, executive function-
ing tasks such as recalling a set of digits in the order in which they
were presented were less effortful and required less self-control
resources than tasks that required the recall of digits in reverse
order. This is because maintaining figures in memory as in the
forward span task does not require other control processes, such as
updating and ignoring or inhibiting competing information, as in
the reverse span task. Similarly, Wright et al. (2008) suggested that
simple math calculations involving single arithmetic operations
(e.g., multiplication) require less effort and place less burden on
self-control resources than mixed-operation calculations, as they
do not require the need to suppress competing processes: “A single
operation challenge . . . obviates the need to inhibit one operational
impulse (e.g., that to add) in favor of another (e.g., that to divide)
and, thus, eliminates a source of regulatory demand” (Wright et al.,
2008, p. 130).
A substantial subset of the present sample of studies (k36)
adopted dependent self-control tasks that varied in their complex-
ity and could be categorized accordingly. These could then be used
as moderator groups to test the hypothesis that simple tasks require
less self-control resources than complex tasks do. This hypothesis
is consistent with the strength model, which suggests that depen-
dent tasks with greater regulatory demand would lead to increased
performance impairment and therefore more ego depletion after
working on a depleting task. Studies that adopted tasks requiring
rote memory, recall, or single arithmetic operations were classified
as simple self-control tasks. These tasks included rote memory
(e.g., Gailliot, Schmeichel, & Baumeister, 2006, Study 7), addition
and multiplication (e.g., Mead, Baumeister, Gino, Schweitzer, &
Ariely, 2009, Study 1; Stewart et al., 2009; Vohs et al., 2005,
Study 1), and forward span recall (e.g., Schmeichel, 2007, Study
2). Tasks that required multiple cognitive processes such as en-
coding, memory maintenance and updating, and multiple arith-
metic operations were classified as complex. These included solv-
able anagrams (e.g., Gailliot, Baumeister, et al., 2007, Study 8;
Gordijn et al., 2004, Study 5); reading, sentence, and reverse span
memory tasks (e.g., Johns et al., 2008, Studies 1 and 4;
Schmeichel, 2007, Studies 1 and 2); Graduate Record Examination
reasoning (e.g., Fennis et al., 2009, Study 2; Finkel et al., 2006,
Studies 2 and 3; Gailliot et al., 2006, Study 7); and mixed-
operation arithmetic calculation tasks (Wright et al., 2008). We
again used an expert judge to corroborate our classification, and
interrater reliability for the classifications was very high (␬⫽.89,
p.001). There were insufficient numbers of studies (k9) in
the cognitive processing sphere for us to conduct a meta-analysis
with depleting task complexity as a moderator.
Source laboratory. The strength model was developed and
extensively tested by Baumeister and colleagues (Baumeister et
al., 1998; Muraven et al., 1998). The model comprised the largest
number of tests of the ego-depletion effect emanating from a single
research group in the present sample of studies. We therefore
considered it prudent to include source laboratory as a moderator
variable to detect any potential variation in the effect in studies
originating in the laboratories of Baumeister and his known col-
laborators (DeWall, Gailliot, Muraven, Schmeichel, and Vohs) and
studies from other investigators’ laboratories. Studies were coded
if they emanated from the Baumeister laboratory or those of his
collaborators (k98) or another laboratory (k100) according
to coauthorship and known prior affiliation.
Moderator Analysis
Analytic strategy. The influence of categorical moderator
variables on the ego-depletion effect was evaluated by segregating
studies on the basis of the moderator and conducting separate
meta-analyses in each moderator group. The moderator was con-
sidered effective if the average corrected effect sizes calculated in
each moderator group were significantly different, as evidenced by
no overlap in the CI
. Moderation was further supported if the
moderator resulted in a narrowing of the CI
, an increase in the
variance accounted for by sampling error variability, and a de-
crease in the I
statistic. One of the moderator variables (depleting
task duration) was continuous, and we tested its effect on the
ego-depletion effect size using linear regression.
Relations between moderators. We also examined relations
between the significant moderator variables to identify potential
confounding effects among the moderators of the ego-depletion
effect (DeCoster, 2004; Lipsey, 2003). As the majority of the
moderators in the present study were categorical, we conducted chi-
square analyses on two-way categorical tables to evaluate the strength
of the relationships between the variables. Relations between the one
continuous moderator (depleting task duration) and the categorical
moderators were evaluated by calculation of the average values for
the continuous moderator at each level of the categorical variable
followed by a one-way ANOVA to test the strength of the rela-
tionship. A statistically nonsignificant chi-square or Fratio indi-
cates that the effect of one moderator variable on the dependent
variable (ego depletion) is likely to be independent of that of the
other moderator variable. We used a conservative pvalue of .01 to
evaluate the significance of the relations. In the event of a signif-
icant association between moderators, we examined the categorical
tables to identify the source of the variation. Where appropriate,
we conducted follow-up meta-analyses of the ego-depletion effect
in groups of studies determined by the crossing of the related
moderator variables.
Additional Dependent Variables and
Alternative Hypotheses
Studies measuring additional dependent variables alongside
measures of task performance in ego-depletion studies adopting
the dual-task paradigm were also subjected to meta-analytic syn-
thesis. In particular, studies including self-report measures of
effort, positive and negative affect, perceived difficulty, subjective
fatigue, self-efficacy, and blood glucose for experimental (ego-
depleted) and control (nondepleted) groups were identified and
effect sizes calculated. These additional variables were not mea-
sured in all studies or were unavailable in some cases, so the
samples for these analyses were invariably subsets of the sample
used to calculate the overall ego-depletion effect.
Positive and negative affect were tapped with previously vali-
dated (e.g., BMIS; Mayer & Gaschke, 1988; PANAS; Watson,
Clark, & Tellegen, 1988) and author-developed measures of mood
and affect. Subjective fatigue and perceived difficulty were invari-
ably measured on self-report scales developed by the study au-
thors. Measures of self-efficacy included standardized measures
(e.g., Generalized Self-Efficacy Scale; Schwarzer & Jerusalem,
1995) as well as those developed by the study authors. A content
analysis of the items used in the self-efficacy instruments revealed
considerable commonality in content with measures focusing on
perceived competency and expectations of success in performing
the target task. A judge employed to independently code the
dependent measures demonstrated strong agreement with our clas-
sification (␬⫽.95, p.001). Discrepancies were resolved using
the method we had used previously. Studies measuring the effect
of ego depletion on blood glucose levels used commercially avail-
able electronic blood glucose analyzers.
Testing Strength Model Hypotheses
Conservation. Studies testing the conservation hypothesis
were required to provide sufficient data for us to calculate a
standardized difference in postdepletion performance on the
second self-control task for participants told to expect an addi-
tional self-control task and control participants not informed of
an additional task. Larger effect sizes reflected greater levels of
ego depletion in participants anticipating future self-control.
We also tested the effect of strategies to increase motivation on
the ego-depletion effect across studies. We therefore meta-
analyzed the second-task performance of ego-depleted partici-
pants provided with either a motivational intervention or no
intervention in the dual-task paradigm. Motivational interven-
tions included providing monetary incentives to complete the
second task (Muraven & Slessareva, 2003), framing the second
task as important or meaningful (Muraven & Slessareva, 2003),
and presenting the second task in an autonomy-supportive man-
ner to facilitate intrinsic motivation (Moller, Deci, & Ryan,
2006; Muraven, Gagne, & Rosman, 2008). Larger effect sizes
indicated that motivation was effective in promoting better
performance on self-control tasks.
Training. The criterion for studies eligible for the training
hypothesis analysis was the inclusion of an experimental manip-
ulation that required participants to engage in a period of practice
with self-control tasks prior to being subjected to an ego-depletion
manipulation using the dual-task paradigm. Self-control was
trained with diverse tasks, such as speech modification (e.g., using
“yes” and “no” only and no colloquialisms; Gailliot, Plant, et al.,
2007), use of the nondominant hand (Finkel, DeWall, Slotter,
Oaten, & Foshee, 2009), maintaining posture, mood regulation,
monitoring eating habits (Muraven et al., 1999), impulse control
tasks like the Stroop test or an aversive mouth rinse (Hui et al.,
2009), physical exercise (Oaten & Cheng, 2006b), financial self-
management (Oaten & Cheng, 2007), and academic study (Oaten
& Cheng, 2006a). Studies reporting differences in second self-
control task performance between trained depleted participants and
untrained depleted controls were included. Larger effect sizes
represented superior performance on self-control tasks as a result
of training.
Recovery. We aimed to test the variation of the ego-depletion
effect in studies that manipulated the duration of the intertask rest
or relaxation period in the dual-task paradigm. However, we could
locate only two studies (k4) that tested this effect (Oaten et al.,
2008; Tyler & Burns, 2008). This number was considered insuf-
ficient for conducting a moderator analysis. An indirect test of the
recovery hypothesis was offered through our analysis of intertask
interim activities reported in the previous section.
Glucose supplementation. Studies that administered glucose-
containing solutions to participants prior to the experiment or in
the interim period between self-control tasks were subjected to a
meta-analysis. The effect size was calculated as the difference in
second task performance for participants assigned to either a group
that received a glucose solution or a control group that received a
sweet placebo. Larger effect sizes were indicative of better per-
formance on self-control tasks as a result of glucose supplemen-
The literature search identified 83 experimental studies that
satisfied inclusion criteria with 198 independent tests of the ego-
depletion effect. Effect sizes for the ego-depletion effect and the
effect of depletion on additional dependent variables (effort, pos-
itive and negative affect, perceived difficulty, subjective fatigue,
and self-efficacy) are reported in Appendix B (see the online
supplemental materials).
The table also includes moderator cod-
ing, a precise description of how the effect size was extracted in
each study, and details of the methods adopted to measure addi-
tional dependent variables. Effect sizes and characteristics of stud-
ies included in the analyses of the conservation (anticipate future
self-control task, motivation) and training hypotheses of the
strength model are provided in Table 1. Analysis of outliers using
Grubbs’ (1950) statistic identified five cases. Three outliers were
detected in the data set for the overall ego-depletion effect, two on
the right-hand side of the distribution (d3.02, Muraven et al.,
2008, Study 1; d2.60, Tyler & Burns, 2009, Study 2, sample 2)
and one on the left-hand side (d⫽⫺0.57, Wright et al., 2007,
Study 2, sample 2). These were replaced with their nearest neigh-
boring values, d1.90 and d⫽⫺0.11, respectively, and retained
in subsequent analyses. In the analyses for additional dependent
variables, one outlier was detected for the negative affect variable
(d1.40, Ciarocco et al., 2001, Study 2) and was also set to its
closest neighboring value (d0.99). In the analyses of additional
hypotheses, one outlier was identified for the conservation hypoth-
esis, expect future task analysis (d2.61, Tyler & Burns, 2009)
and was replaced with its nearest neighbor (d1.70).
Overall Ego-Depletion Effect
Results of the meta-analysis of the overall ego-depletion effect
are presented in Table 2. The averaged corrected standardized
mean difference for ego depletion on self-control dependent mea-
sures was d
0.62, CI
[0.57, 0.67], Q(197) 301.79, p
.001. This represents a medium-to-large effect size (Cohen, 1987).
All but two of the effect sizes in the data set were positive in
valence. The Qstatistic, percentage of variance accounted for by
sampling error variance (65.61%), and an I
statistic exceeding
25% (34.72%) showed a substantial degree of heterogeneity in the
effect size across the studies, which indicated the likelihood of
extraneous moderators of the effect (Higgins & Thompson, 2002).
The confidence intervals did not include zero, leading to a rejec-
tion of the null hypothesis. The fail-safe sample size (N
ceeded Rosenberg’s (2005) cutoff value, indicating that it was
highly unlikely that sufficient studies with null effects would exist
to reduce the ego-depletion effect to a trivial value.
Additional Dependent Variables
Average corrected effect size statistics for the effect of ego
depletion on other dependent variables are given in Table 2. Each
Five effect sizes were included in the analysis testing the effect of ego
depletion on blood glucose. The effect sizes are not reported in Appendix
B, as few of the studies involved in the blood glucose analysis were
included in the meta-analysis of the overall ego-depletion effect. The effect
sizes were from Studies 3 (d1.64, n16), 4 (d1.23, n12), 5 (d
0.96, n23), and 6 (d0.89, n17) in Gailliot et al.’s (2007) article
and Dvorak and Simons’ (2009) study (d0.69, n90).
effect represents the standardized mean difference in the target
dependent variable for ego-depleted participants and nondepleted
controls. Effect sizes for effort, d
0.64, CI
[0.47, 0.80],
Q(30) 95.04, p.001, perceived difficulty, d
0.94, CI
[0.73, 1.14], Q(57) 456.57, p.001, subjective fatigue, d
0.44, CI
[0.26, 0.63], Q(25) 97.61, p.001, and blood
glucose levels, d
⫽⫺0.87, CI
[1.20, 0.54], Q(4) 2.83,
p.59, were significantly different from zero and were medium
to large in magnitude. Effect sizes for positive affect, d
[0.12, 0.05], Q(66) 146.09, p.01, negative affect,
0.14, CI
[0.06, 0.22], Q(35) 45.73, p.11, and
self-efficacy, d
0.16, CI
[0.19, 0.51], Q(4) 8.10, p
.09, were substantially smaller. The confidence intervals for the
positive affect and self-efficacy effect sizes included the value of
Table 1
Characteristics of Studies Used in Meta-Analyses Testing Additional Hypotheses
Hypothesis Author Study Nd
Future self-control Muraven, Shmueli, & Burkley (2006) 1 46 0.91
2 34 1.04
3 31 0.66
4 38 0.61
Tyler & Burns (2009) 1 41 2.53
2 30 1.70
3 40 1.00
Motivation Moller et al. (2006) 1 24 1.05
2 24 3.75
3 30 3.84
Muraven et al. (2008) 1 16 4.59
2 44 0.65
3 48 0.78
Muraven & Slessareva (2003) 1 43 0.60
2 41 0.92
3 24 0.89
Stewart et al. (2009) 42 0.72
Training Finkel et al. (2009) 5 40 0.82
Gailliot, Plant, et al. (2007) 1 20 0.99
2 22 0.91
4 52 0.50
Hui et al. (2009) 42 0.48
Muraven et al. (1999) 69 0.57
Oaten & Cheng (2006a) 45 8.59
Oaten & Cheng (2006b) 24 3.13
Oaten & Cheng (2007) 49 5.97
Glucose supplementation DeWall et al. (2008) 2 30 0.96
Gailliot et al. (2007) 8 73 0.55
9 18 1.57
Gailliot et al. (2009) 51 0.69
Masicampo & Baumeister (2008) 57 0.77
Ego-depletion effect from dual-task paradigm for depleted participant group in the presence and absence of moderator (future self-control, motivation,
training, and glucose supplementation).
Expect future self-control task vs. expect no future self-control task conditions.
Expect future self-control
task vs. expect future hard task conditions.
Average effect size for 3-min and 10-min actual task duration vs. 20-min actual task duration.
finished vs. experiment unfinished conditions.
Autonomy vs. controlled motivational conditions.
Pressure vs. no pressure conditions.
vs. no motivation condition; motivational manipulations included increasing interest in the final task, highlighting the benefits of practice, and providing
monetary incentives.
High expectancy of reward vs. low expectancy of reward conditions.
Effect size is overall effect of regulation training between
self-control exercise vs. control groups; self-control training required participants to engage in either verbal regulation (avoid colloquialisms, slang,
abbreviated, and shorthand terms) or physical regulation (use of nondominant hand for everyday tasks) over a period of 2 weeks.
Self-control exercise
vs. no exercise condition; self-control exercises required modifying speech, modifying posture, controlling emotions, monitoring eating habits, and avoiding
use of colloquialisms over a period of 2 weeks.
Effect size is average across crossing-out-letter and cold-pressor dependent tasks between the strong
training vs. no training groups; self-control exercises required participants to engage in an incongruent Stroop task and rinse their mouth with Listerine for
30 s twice a day over a period of 2 weeks.
Self-control exercise (Cohort 1) vs. no exercise (Cohort 2) groups; self-control exercises required participants
to engage in a regular program of study including self-imposed deadlines, a study schedule, and use of a study register and diary to enhance self-monitoring
over a period of 2 months.
Self-control exercise (Cohort 1) vs. no exercise (Cohorts 2 and 3) groups; self-control exercises required participants to
engage in a regular tailored program of physical activity prescribed by a gym instructor over a period of 2 months.
Self-control exercise (Cohort 1) vs.
no exercise (Cohort 2) groups; self-control exercises required participants to engage in a personal financial management program including spending diaries
and program logs over a period of 4 months.
Effect size is average effect of supplementation (glucose vs. placebo) condition on word fragments task
performance for participants in the high depletion (mortality salience) condition.
Effect size is average effect of supplementation (glucose vs. placebo)
condition on self-reported willingness to help after participants had completed an examination as part of their psychology course.
Effect size is average
effect of supplementation (glucose vs. placebo) condition on number of voluntary helping hours given for participants in the high depletion (attention
control) condition.
Effect size is average effect of supplementation (glucose vs. placebo) condition on number of prejudiced statements given for
participants high in prejudiced attitudes.
Effect size is average effect of supplementation (glucose vs. placebo) condition on choice of apartment
(attraction effect) for participants in the high depletion (attention control) condition.
zero. With the exception of effect sizes for negative affect and
blood glucose, all effect sizes exhibited significant heterogeneity.
Moderator Analyses
Averaged corrected standardized-difference effect sizes and
associated statistics for each moderator group are provided in
Table 3.
Sphere of depleting task. We tested the specific and general
spheres of depleting task as moderators of the overall ego-depletion
effect. Medium-to-large effect sizes were observed for moderator
groups defined according to the specific sphere of depleting task
adopted: controlling emotions, d
0.62, CI
[0.50, 0.74], Q(23)
33.67, p.07, thoughts, d
0.63, CI
[0.53, 0.75], Q(24)
18.16, p.80, impulses, d
0.55, CI
[0.46, 0.64], Q(60)
128.26, p.001, and attention, d
0.65, CI
[0.50, 0.81],
Q(20) 10.68, p.95; choice and volition, d
0.82, CI
1.00], Q(11) 8.87, p.63; and cognitive, d
0.54, CI
0.80], Q(12) 30.64, p.01, and social, d
0.75, CI
0.84], Q(37) 30.69, p.76, processing tasks. Five of these effects
were homogenous. Confidence intervals revealed that the ego-
depletion effect size for the controlling impulses task moderator
group was significantly smaller than the effect size for the social
processing task moderator group. There were no other significant
differences. Moderator analysis of experiments adopting affective,
0.57, CI
[0.49, 0.64], Q(87) 163.59, p.001, cognitive,
0.65, CI
[0.57, 0.73], Q(70) 95.23, p.05, and
combined affective and cognitive, d
0.73, CI
[0.64, 0.83],
Q(38) 32.56, p.67, depleting tasks produced effect sizes of
similar magnitude. There were no significant differences in effect
sizes across the moderator groups. Only the effect size for studies
classified as combined affective and cognitive was homogeneous.
Frequently used depleting tasks. We also tested whether the
average ego-depletion effect size varied across studies adopting
frequently used depleting tasks. Analyses for crossing out letters,
0.77, CI
[0.65, 0.90], Q(19) 25.68, p.14, Wegner’s
white-bear paradigm, d
0.65, CI
[0.52, 0.78], Q(18) 16.72,
p.54, video-watching affect regulation, d
0.55, CI
0.68], Q(18) 18.83, p.40, and video-watching attention
control, d
0.61, CI
[0.48, 0.74], Q(18) 23.89, p.16,
tasks revealed medium-to-large average effect sizes for these tasks
with no significant differences across moderator groups. The effect
size for the modified Stroop task was smaller in magnitude, d
0.40, CI
[0.26, 0.55], Q(12) 23.01, p.05, and significantly
smaller than the effect size for the crossing-out-letters task. With
the exception of the Stroop task group, all of the moderator
analyses yielded homogeneous effect sizes.
Duration of depleting task. Depleting task duration in min-
utes was treated as a continuous moderator of the ego-depletion
effect. The majority of studies (k148) reported task duration
data, and this was converted into minutes. We conducted a linear
regression analysis, with the ego-depletion effect size as the
The modified Stroop task was the only specific self-control depleting
task that exhibited significant heterogeneity and a mean effect size signif-
icantly lower than those in the other depleting task moderator groups. This
pattern mirrored the pattern of effect sizes for tasks in the controlling
impulse and affective groups in the spheres of depleting task moderator
analysis. We therefore tested whether studies adopting the Stroop task were
responsible for the heterogeneity and significantly lower averaged effect
sizes observed in these spheres by omitting studies adopting the Stroop task
(k13) from the controlling impulse and affective moderator analyses.
Reestimating the meta-analysis revealed that omitting these tasks did not
lead to a substantive change in the averaged effect size or an appreciable
narrowing of the confidence intervals for the controlling impulses sphere,
0.62, CI
[0.51, 0.74], Q(47) 95.16, p. 001, or the affective
sphere, d
0.62, CI
[0.54, 0.70], Q(74) 129.47, p. 001, of the
depleting task. The substantial heterogeneity remained.
Table 2
Results of Meta-Analysis of Ego Depletion
Effect kN d
Var QI
Overall ego-depletion effect 198 10,782 0.62 0.57 0.67 0.20 0.02 50,445 65.61 301.79
34.72 22.14 45.27
Other dependent variables
Effort 31 1,835 0.64 0.47 0.80 0.39 0.09 1,383 32.62 95.04
69.49 55.69 78.99
Positive affect 67 4,033 0.03 0.12 0.05 0.28 0.05 0 45.86 146.09
54.82 40.48 65.71
Negative affect 36 2,237 0.14 0.06 0.22 0.13 0.04 55 78.72 45.73 23.46 0.00 49.56
Perceived difficulty 58 3,597 0.94 0.73 1.14 0.72 0.10 12,994 12.70 456.57
87.52 84.62 89.86
Subjective fatigue 26 1,809 0.44 0.26 0.63 0.41 0.09 554 26.64 97.61
74.39 62.44 82.54
Self-efficacy 5 210 0.16 0.19 0.51 0.25 0.18 0 61.76 8.10 50.62 0.00 81.88
Blood glucose 5 158 0.87 1.20 0.54 0.00 0.17 28 100.00 2.83 0.00
Strength model hypotheses
Conservation hypothesis
Future self-control 7 260 1.04 0.78 1.30 0.15 0.13 106 83.28 8.41 28.66 0.00 69.31
Motivation 10 336 1.05 0.49 1.61 0.83 0.29 301 17.82 56.12
83.96 72.08 90.79
Training hypothesis 9 363 1.07 0.10 2.03 1.43 0.49 567 6.48 138.98
94.24 91.09 96.28
Glucose supplementation 5 229 0.75 0.48 1.03 0.00 0.14 32 100.00 3.36 0.00
Note.kno. effect sizes in meta-analysis; Ntotal sample size in meta-analysis; d
averaged corrected standardized difference effect size; CI
95% confidence intervals; LL lower limit of confidence interval; UL upper limit of confidence interval; SD residual standard deviation of d
standard error of d
fail-safe N; Var % variance attributed to sampling error variability; QCochran’s (1952) Qstatistic; I
Higgins and
Thompson’s (2002) I
Table 3
Results of Moderator Analyses
Moderator kNd
Var QI
Sphere of depleting task
Specific task classification
Controlling emotions 24 1,146 0.62 0.50 0.74 0.19 0.06 636 71.28 33.67 34.69 0.00 58.51
Controlling thoughts 25 1,212 0.63 0.53 0.75 0.00 0.06 708 100.00 18.16 0.00
Controlling impulses 61 3,956 0.55 0.46 0.64 0.27 0.05 4,434 47.56 128.26
53.22 37.39 65.05
Controlling attention 21 1,078 0.65 0.50 0.81 0.22 0.08 569 63.80 10.68 0.00
Choice and volition 12 475 0.82 0.62 1.00 0.00 0.10 213 100.00 8.87 0.00
Cognitive processing 13 615 0.54 0.29 0.80 0.35 0.13 136 42.42 30.64
60.84 28.21 78.63
Social processing 38 1,966 0.75 0.65 0.84 0.00 0.05 1,471 100.00 30.69 0.00
Global classification
Affective 88 5,351 0.57 0.49 0.64 0.24 0.04 9,086 53.79 163.59
46.82 31.66 58.61
Cognitive 71 3,380 0.65 0.57 0.73 0.18 0.04 6,237 74.56 95.23
26.49 0.78 45.55
Affective and cognitive 39 2,051 0.73 0.64 0.83 0.00 0.05 2,512 100.00 32.56 0.00
Frequently used depleting
Crossing out letters 20 1,101 0.77 0.65 0.90 0.15 0.06 760 77.88 25.68 26.01 0.00 57.09
White bear paradigm 19 932 0.65 0.52 0.78 0.00 0.07 432 100.00 16.72 0.00
Video-watching affect
19 963 0.55 0.42 0.68 0.00 0.07 315 100.00 18.83 0.00
Video-watching attention
19 1,006 0.61 0.48 0.74 0.14 0.06 411 79.54 23.89 24.65 0.00 56.83
Modified Stroop 13 1,322 0.40 0.26 0.55 0.18 0.07 160 56.50 23.01
47.85 72.55 0.91
Interim period
Completed questionnaires 119 5,726 0.71 0.65 0.77 0.18 0.03 20,948 73.47 161.96
27.14 8.06 42.26
Filler task and/or break
between tasks 24 1,115 0.72 0.60 0.85 0.00 0.06 6,662 100.00 22.30 0.00
No interim period reported 55 3,941 0.47 0.39 0.55 0.17 0.04 2,876 66.11 83.19
35.08 9.57 53.40
Experiment presentation
Presented as single
experiment 134 7,413 0.58 0.53 0.64 0.19 0.03 20,455 68.40 195.91
32.11 15.75 45.30
Presented as separate
experiment 64 3,369 0.71 0.63 0.80 0.21 0.04 6,627 66.12 96.80
34.92 11.45 52.17
Administered by same
experimenter 183 10,133 0.61 0.56 0.66 0.21 0.03 41,876 64.54 283.55
35.81 22.96 46.52
Administered by different
experimenters 15 649 0.86 0.69 1.01 0.00 0.07 404 100.00 9.76 0.00
Sphere of dependent task
Specific classification
Controlling impulses 104 4,836 0.71 0.64 0.78 0.21 0.03 15,427 68.84 151.07
31.82 12.89 46.63
Choice and volition 8 1,300 0.22 0.11 0.33 0.00 0.06 23 100.00 4.19 0.00
Cognitive processing 47 2,505 0.60 0.52 0.67 0.00 0.04 2,470 100.00 43.03 0.00
Social processing 33 1,934 0.69 0.60 0.78 0.00 0.05 1,797 100.00 29.11 0.00
Global classification
Affective 107 4,945 0.71 0.64 0.78 0.21 0.04 16,366 68.72 155.70
31.92 13.33 46.53
Cognitive 58 3,903 0.48 0.40 0.55 0.17 0.04 3,106 67.66 85.73
33.51 8.01 51.94
Affective and cognitive 33 1,934 0.69 0.60 0.78 0.00 0.05 1,797 100.00 29.11 0.00
Frequently used dependent
Handgrip 18 872 0.64 0.45 0.83 0.28 0.10 407 53.18 33.85
49.78 13.52 70.84
Modified Stroop 15 633 0.76 0.59 0.94 0.16 0.09 306 80.88 18.55 24.53 0.00 59.25
Food taste test 10 527 0.50 0.33 0.68 0.16 0.09 71 75.71 13.21 31.87 0.00 67.48
Math or mental arithmetic 10 447 0.50 0.31 0.69 0.00 0.10 55 100.00 7.80 0.00
Solvable anagram 10 670 0.60 0.44 0.76 0.08 0.08 128 91.43 10.94 17.73 0.00 58.74
Matched depleting and
dependent tasks
Specific classification
Matched 40 1,881 0.59 0.48 0.71 0.23 0.06 1,565 63.60 62.90
38.00 8.99 57.76
Unmatched 158 8,901 0.63 0.58 0.68 0.20 0.03 33,925 66.25 238.49
34.17 19.79 45.97
Global classification
Matched 68 3,226 0.61 0.52 0.70 0.21 0.04 5,067 67.92 100.12
33.08 9.60 50.46
Unmatched 130 7,556 0.63 0.57 0.69 0.20 0.03 23,300 64.51 201.52
35.99 20.54 48.43
(table continues)
dependent variable and task duration as an independent predic-
tor, weighted by the inverse variance of each effect size
(Hedges & Olkin, 1983; Sa´nchez-Meca & Marı´n-Martı´nez,
1998). The analysis yielded a marginally significant effect for
task duration on ego depletion, but the effect size was small
(␤⫽.11, z1.79, p.07).
Interim period. We conducted an analysis examining the
effect of the type of interim activity on ego depletion in the
dual-task paradigm. The averaged effect size did not vary across
moderator groups for studies in which participants completed
questionnaires during the interim period, d
0.71, CI
0.77], Q(118) 161.96, p.01, or completed a filler task or had
a break between the tasks, d
0.72, CI
[0.60, 0.85], Q(23)
22.30, p.50, but the effect for studies that reported no interim
period was significantly smaller, d
0.47, CI
[0.39, 0.55],
Q(54) 83.19, p.01. The analysis for the filler task or break
group was homogeneous, but the remaining effects exhibited sub-
stantial heterogeneity.
Experiment presentation. The ego-depletion effect for stud-
ies presenting the two tasks as single experiments, d
0.58, CI
[0.53, 0.64], Q(133) 195.91, p.001, or separate experiments,
0.71, CI
[0.63, 0.80], Q(63) 96.80, p.001, were not
significantly different. However, the effect size for tasks adminis-
tered by the same experimenter, d
0.61, CI
[0.56, 0.66],
Q(182) 283.55, p.001, was significantly smaller than that for
tasks administered by a different experimenter, d
0.86, CI
[0.69, 1.01], Q(14) 9.76, p.78. Effect sizes were medium to
large in all cases, and all effects with the exception of the effect for
tasks administered by different experimenters exhibited substantial
Sphere of dependent task. We tested whether groups of
effect sizes defined according to the specific features of the crite-
rion self-control tasks would moderate the ego-depletion effect.
Three of the moderator groups (controlling attention, controlling
emotions, and controlling thoughts) contained fewer than five
effect sizes and were deemed insufficient for meta-analytic syn-
thesis. Analyses of the four remaining moderator groups revealed
no significant differences in effect sizes for tasks in the controlling
impulses sphere, d
0.71, CI
[0.64, 0.78], Q(103) 151.07,
p.01, cognitive processing sphere, d
0.60, CI
[0.52, 0.67],
Q(46) 43.03, p.60, and social processing sphere, d
[0.60, 0.79], Q(32) 29.11, p.61. However, confidence
intervals indicated that the ego-depletion effect was significantly
smaller in experiments adopting choice and volition tasks, d
0.22, CI
[0.11, 0.33], Q(7) 4.19, p.76, than in all other
groups. All of the effect sizes were homogeneous, with the excep-
tion of the effect in the controlling impulses tasks group. For the
moderator analysis in which dependent self-control tasks were
classified in accordance with their global features, effect sizes for
affective tasks, d
0.71, CI
[0.64, 0.78], Q(106) 155.70,
p.01, and combined affective and cognitive tasks, d
[0.60, 0.79], Q(32) 29.11, p.61, were significantly
larger relative to the effect for cognitive tasks, d
0.48, CI
[0.40, 0.55], Q(57) 85.73, p.01. Only the effect size for the
combined affective and cognitive group was homogeneous.
Table 3 (continued)
Moderator kNd
Var QI
Type of dependent task
Behavioral self-control measure 170 9,096 0.62 0.56 0.67 0.22 0.03 35,689 61.91 274.61
38.46 25.77 48.97
Nonbehavioral judgment measure 28 1,686 0.66 0.56 0.76 0.00 0.05 1,231 100.00 26.44 0.00
Control condition task
Easier version 170 9,272 0.62 0.56 0.67 0.23 0.03 36,526 60.03 283.19
40.32 28.13 50.44
Individual difference 21 1,194 0.66 0.54 0.78 0.00 0.06 618 100.00 11.37 0.00
Alternative 5 274 0.53 0.29 0.77 0.00 0.12 16 100.00 0.23 0.00
Task complexity
Complex 27 1,495 0.65 0.54 0.75 0.00 0.05 972 100.00 26.44 0.00
Simple 9 407 0.35 0.15 0.55 0.00 0.10 16 100.00 5.93 0.00
Source laboratory
Baumeister 98 5,906 0.60 0.53 0.66 0.21 0.03 12,425 61.99 158.08
38.64 21.51 52.02
Other 100 4,876 0.66 0.59 0.72 0.19 0.04 12,758 70.71 141.43
30.00 10.01 45.55
Note.kno. effect sizes in meta-analysis; Ntotal sample size in meta-analysis; d
averaged corrected standardized difference effect size; CI
95% confidence intervals; LL lower limit of confidence interval; UL upper limit of confidence interval; SD residual standard deviation of d
standard error of d
fail-safe N; Var % variance attributed to sampling error variability; QCochran’s (1952) Qstatistic; I
Higgins and
Thompson’s (2002) I
Two studies (k4) were omitted from the specific classification analysis because two depleting self-control tasks in different spheres were used
concurrently (Fischer et al., 2007, Studies 1, 2, and 4; Ostafin, Marlatt, & Greenwald, 2008).
The studies that used depleting tasks in two spheres were
classified into the affective (Fischer et al., 2007, Studies 1, 2, and 4) or combined affective and cognitive (Ostafin et al., 2008) moderator groups for the
global classification analysis.
One study (k3) was omitted from this analysis because the emotion regulation tasks were used in conjunction with tasks
in another sphere (Fischer et al., 2007, Studies 1, 2, and 4).
One study (k3) was omitted from this analysis because the attention control task was
used in conjunction with a task in another sphere (Fischer et al., 2007, Studies 1, 2, and 4).
Three moderator groups had four or fewer effect sizes and
were deemed unsuitable to be included in the analysis: controlling emotions (k3; Muraven et al., 1998, Study 3; Schmeichel, 2007, Study 3; Vohs et
al., 2005, Study 3), controlling attention (k3; Muraven et al., 2008, Studies 1 and 3; Muraven, Shmueli, & Burkley, 2006, Study 2), and controlling
thoughts (k0).
Only two effect sizes were available for the moderator group in which participants allocated to the control condition did not engage
in any task (Baumeister et al., 1998, Study 2; Burkley, 2008, Study 2), so this moderator group was omitted from the analysis.
Frequently used dependent tasks. Five dependent self-control
tasks were used on 10 or more occasions. Averaged effect sizes for
handgrip, d
0.64, CI
[0.45, 0.83], Q(17) 33.85, p.01,
modified Stroop, d
0.76, CI
[0.59, 0.94], Q(14) 18.55, p
.18, food taste test, d
0.50, CI
[0.33, 0.68], Q(9) 13.21, p
.20, math or mental arithmetic, d
0.50, CI
[0.31, 0.69],
Q(9) 7.80, p.55, and solvable anagram, d
0.60, CI
0.76], Q(9) 10.94, p.28, were medium to large in size and were
not significantly different across tasks. With the exception of the
handgrip task, all of the moderator analyses resulted in homogeneous
effect sizes.
Matched depleting and dependent tasks. The ego-depletion
effect size for studies that adopted depleting and dependent tasks
that were matched on task sphere, specific classification, d
0.59, CI
[0.48, 0.71], Q(39) 62.90, p.01; global classifi-
cation, d
0.61, CI
[0.52, 0.70], Q(67) 100.12, p.001,
did not differ from the effect size for studies adopting tasks
unmatched on sphere: specific classification, d
0.63, CI
[0.58, 0.68], Q(157) 238.49, p.001; global classification,
0.63, CI
[0.57, 0.69], Q(129) 201.52, p.001.
the moderator group in which the tasks were matched according to
specific sphere classification was homogeneous.
Type of dependent task. Tests of the ego-depletion effect for
studies employing a behavioral dependent measure, d
[0.56, 0.67], Q(169) 274.61, p.001, or a nonbehavioral
or judgment dependent measure, d
0.66, CI
[0.56, 0.76],
Q(27) 26.44, p.49, revealed medium-to-large effect sizes in
each moderator group with no significant differences. The effect
size for behavioral measures exhibited substantial heterogeneity,
but the effect size for nonbehavioral or judgment measures was
Control condition task. Meta-analyses of studies using an
easier version of the depleting task, d
0.62, CI
[0.56, 0.67],
Q(169) 283.19, p.001, an alternative task, d
0.53, CI
[0.29, 0.77], Q(4) 0.23, p.99, and the same task with an
individual difference characteristic to evoke depletion, d
[0.54, 0.78], Q(20) 11.37, p.94, revealed no differences
in effect size. The effect size for studies using an easy version of
the depleting task exhibited substantial heterogeneity, but the
effect sizes for the alternative task and same task with individual
differences groups were homogeneous.
Task complexity. Ego-depletion effect sizes were larger for
the subset of studies adopting complex dependent tasks, d
0.65, CI
[0.54, 0.75], Q(26) 26.44, p.44, than for those
adopting less complex or simple tasks, d
0.35, CI
0.55], Q(8) 5.93, p.66, a difference that was borderline
significant. In both cases, the averaged effect size was homoge-
Source laboratory. There were no significant differences in
the average effect size for studies originating from the labs of
Baumeister and collaborators, d
0.60, CI
[0.53, 0.66],
Q(97) 158.08, p.001, and from other laboratories, d
[0.59, 0.72], Q(99) 141.43, p.01. Both effects exhibited
substantial heterogeneity.
Individual differences. Several studies (k9) measured trait
self-control concurrent with ego depletion in dual-task paradigm
experiments (DeWall et al., 2007, Study 4; Dvorak & Simons,
2009; Finkel & Campbell, 2001, Study 2; Gailliot & Baumeister,
2007b, Studies 1 and 2; Gailliot, Schmeichel, & Maner, 2007,
Studies1&2;Muraven, Pogarsky, & Shmueli, 2006; Stillman et
al., 2009, Study 2). However, the studies did not provide sufficient
data to test individual differences in self-control as a moderator of
the ego-depletion effect in our meta-analysis. In particular, effect
size data for the ego-depletion effect in high and low trait self-
control groups were not available. No other individual difference
or personality constructs were measured with sufficient regularity
to offer a cumulative analysis of their effects on ego depletion.
Relations between moderators. Pairwise tests of relations
between significant moderator variables are presented in Appendix
C (see online supplemental materials). The variables included in
the analysis were sphere of depleting task (specific classification),
frequently used depleting task, experiment presentation (same vs.
different experimenters), interim period, sphere of depleting task
(specific and global classifications), and task complexity. The
analysis yielded relatively few significant relations among the
variables, suggesting that our selection of moderators and coding
procedures were largely effective in identifying unique moderator
variables that were not confounded with others (Lipsey, 2003).
There were significant associations between sphere of depleting
task (specific classification) and frequently used depleting task
moderator variables,
(12) 270.00, p.001, and between
sphere of dependent task, specific and global classifications,
(10) 396.00, p.001. These findings were unsurprising, as
the variables involved represented nested versions of the same
analysis according to our moderator coding. For example, the
sphere of dependent task moderator variables, classified according
to specific and global features, were strongly associated because
specific domains of depleting task (e.g., controlling impulses or
controlling emotions) were exclusive to categories of depleting
tasks in the global domain (e.g., affective tasks). Conducting
follow-up analyses on the basis of these relations was not war-
ranted, as such analyses would fail to explain additional variance
in the ego-depletion effect across moderator groups.
We found significant relations between the interim period and
sphere of depleting task (specific classification) moderators,
(12) 36.39, p.001. Examination of the classification tables
identified that the relationship was due to a disproportionate num-
ber of cases in the “no interim reported” category that adopted
social processing and controlling impulses tasks relative to the
other categories. As our moderator analysis revealed that studies
adopting no interim period between tasks exhibited a significantly
smaller ego-depletion effect than studies that included an interim
period/filler task or had participants complete questionnaires, we
conducted follow-up analyses to investigate whether task sphere
explained this variance. We conducted separate meta-analyses in
We also examined relations among the specific and global task sphere
and frequently used tasks moderator variables across depleting and depen-
dent task types. This was to evaluate whether there was a tendency for
experimenters to adopt tasks from the same sphere for the depleting and
dependent tasks in the dual-task paradigm. We found no association for any
combination of these variables pointing to the independence of these
moderators. This was consistent with the null finding in the moderator
analysis, in which depleting and dependent tasks were coded as matched on
task sphere. Taken together, these results indicate that experimenters in the
present sample of studies tended to adopt depleting and dependent tasks
from different spheres of self-control and demonstrate their utility in
testing the generality of the ego-depletion effect.
moderator groups defined by depleting task sphere (social process-
ing vs. controlling impulses vs. other spheres) and interim period
(interim period/filler task/completed questionnaires vs. no interim
reported). Results indicated that the ego-depletion effect was sig-
nificantly larger in several moderator groups—social processing
tasks–interim period, d
0.79, CI
[0.68, 0.92], Q(23) 19.51,
p.67, social processing tasks–no interim reported, d
[0.54, 0.82], Q(13) 9.71, p.72, controlling impulses–
interim period, d
0.67, CI
[0.56, 0.78], Q(45) 77.40, p
.01, and other spheres–interim period, d
0.71, CI
0.78], Q(70) 73.30, p.37—relative to the effect in the tasks
from the controlling impulses–no interim reported moderator
group, d
0.37, CI
[0.23, 0.52], Q(14) 30.53, p.01.
Furthermore, the effect size in the other studies–no interim re-
ported moderator group, d
0.49, CI
[0.34, 0.65], Q(23)
46.48, p.01, exhibited significant overlap in confidence inter-
vals with effect sizes in the other moderator groups, with the
exception of the social processing tasks–interim period group.
These results suggest that the inclusion of studies adopting con-
trolling impulses depleting tasks led to the significant attenuation
of the ego-depletion effect in the no interim reported moderator
group. However, as the confidence intervals for the other
studies–no interim reported moderator group straddled those of the
controlling impulses–no interim reported moderator group and
those from other moderator groups, it is likely that the association
among these moderators did not fully account for the variation in
the ego-depletion effect size due to interim period.
The presentation of tasks by the same or different experimenters
significantly moderated the ego-depletion effect. This moderator
variable was also significantly related to sphere of depleting task
(specific classification),
(6) 20.40, p.01. We therefore
investigated whether depleting task sphere confounded this effect.
An examination of the classification tables for these variables
revealed that a disproportionately larger proportion of studies
adopting depleting tasks in the social processing sphere also used
different experimenters. For many of the studies adopting deplet-
ing tasks in the social processing sphere, the use of different
experimenters was an integral part of the depletion paradigm (e.g.,
experiments requiring an interracial interaction; Richeson & Shel-
ton, 2003; Richeson & Trawalter, 2005; Richeson et al., 2005;
Trawalter & Richeson, 2006). A follow-up moderator analysis by
depleting task sphere (social processing vs. other spheres) and
experiment presentation (same vs. different experimenters) re-
vealed that studies adopting depleting tasks in the social process-
ing sphere and used different experimenters exhibited a larger
ego-depletion effect, d
1.00, CI
[0.75, 1.24], Q(7) 3.93,
p.79, than studies adopting social processing tasks and using
the same experimenter, d
0.71, CI
[0.60, 0.80], Q(29)
22.08, p.82. Similarly, studies adopting tasks from other
spheres and using different experimenters exhibited a larger ego-
depletion effect, d
0.76, CI
[0.56, 0.97], Q(6) 6.09, p
.41, than studies adopting tasks from other spheres and using the
same experimenter, d
0.59, CI
[0.53, 0.65], Q(148)
253.18, p.001. The ego-depletion effect was significantly larger
in the social processing tasks–different experimenter moderator
group than in the other tasks–same experimenter moderator group.
However, within the spheres of task moderator groups, there were
no significant differences in the effect sizes, and the same pattern
of findings to the main moderator analysis for experiment presen-
tation was observed.
Testing Strength-Model Hypotheses
Averaged corrected standardized-difference effect sizes and as-
sociated statistics for the meta-analyses of additional strength
model hypotheses are provided in Table 2.
Conservation. We meta-analyzed studies in which partici-
pants were informed they were required to engage in a third
self-control task after the second task in the dual-task paradigm.
Two studies were included in the analysis with seven separate tests
of the effect (Muraven, Shmueli, & Burkley, 2006; Tyler & Burns,
2009). The effect size was large, significant, and homogeneous,
1.04, CI
[0.78, 1.30], Q(6) 8.41, p.21, such that
expecting future self-control performance led to higher levels of
ego depletion among depleted samples.
We also tested the moderation of the ego-depletion effect by
motivational strategies. Three studies were included in the analysis
offering 10 independent tests of the effect (Moller et al., 2006;
Muraven et al., 2008; Muraven & Slessareva, 2003). The analysis
produced a large effect size, d
1.05, CI
[0.49, 1.61], Q(9)
56.12, p.001, supporting the hypothesis that introducing moti-
vational strategies results in better performance on dependent
self-control tasks among ego-depleted people. The effect exhibited
significant heterogeneity.
Training. Seven studies provided nine tests of the training
hypothesis (Finkel et al., 2009; Gailliot, Plant, et al., 2007; Hui et
al., 2009; Muraven et al., 1999; Oaten & Cheng, 2006a, 2006b,
2007). The effect size was large and significant, indicating that
participants receiving training performed better on the self-control
task and were less ego-depleted than untrained participants, d
1.07, CI
[0.10, 2.03], Q(8) 138.98, p.001. It is important to
note that none of the individual effects included in the analysis
were small or negative in valence, and the high degree of hetero-
geneity in the effect reflected the very large range in effect sizes
(d0.10 to 8.59).
Glucose supplementation. We tested the effect of adminis-
tering glucose (as opposed to placebo) on the ego-depletion effect.
Four studies reporting five separate effect sizes were included in
the analysis (DeWall et al., 2008; Gailliot, Baumeister, et al., 2007;
Gailliot et al., 2009; Masicampo & Baumeister, 2008). Study
characteristics are reported in Table 1. Results revealed a large
homogeneous effect size, d
0.75, CI
[0.48, 1.03], Q(4)
3.36, p.50, indicating that supplementing people with glucose
is associated with significantly better performance on self-control
tasks among depleted people relative to controls provided with a
sweet placebo.
Our main aim in the present study was to conduct a meta-
analytic synthesis of the ego-depletion effect in experimental stud-
ies adopting the dual-task paradigm. Additional aims were to test
(a) the effect of ego depletion on other dependent variables (effort,
perceived difficulty, subjective fatigue, negative and positive af-
fect, self-efficacy, and blood glucose), (b) the effect of
theoretically-salient moderators on ego depletion, (c) alternative
explanations for the ego-depletion effect, and (d) key additional
hypotheses derived from the strength model of self-control. A
literature search identified 83 studies that met inclusion criteria
and provided 198 independent tests of the ego-depletion effect.
The meta-analysis of the overall ego-depletion effect produced a
medium-to-large average effect size. The effect was significantly
different from zero but displayed substantial heterogeneity. Sig-
nificant effects were found for ego depletion on effort, perceived
difficulty, negative affect, subjective fatigue, and blood glucose,
reflecting the effortful, aversive nature of self-control tasks. The
effect of ego depletion on positive affect and self-efficacy was
trivial and nonsignificant. Moderator analyses revealed that the
ego-depletion effect was generally consistent across sphere of
depleting and dependent task. The effect size for depleting tasks in
the controlling impulses sphere and for the modified Stroop task
was smaller relative to a minority of other spheres and tasks. The
effect was significantly smaller for studies adopting dependent
tasks in the cognitive and choice and volition spheres. The effect
was also smaller for simple as opposed to complex dependent
tasks, a difference that was borderline significant; for tasks admin-
istered by the same experimenter compared to those administered
by different experimenters; and for studies that reported no inter-
task interim period. Presentation of the tasks as separate experi-
ments, whether or not depleting and dependent tasks were matched
on sphere, use of behavioral or nonbehavioral dependent measures,
type of control condition, and source laboratory, did not moderate
the effect. Tests of the additional hypotheses of the strength model
revealed that motivational incentives, training on self-control
tasks, and glucose supplementation resulted in significantly supe-
rior second task performance relative to that of controls among
ego-depleted groups. In contrast, participants who anticipated a
future self-control task exhibited a significantly larger ego-
depletion effect than participants who did not expect a future task
in ego-depleted groups.
Implications for the Strength Model
The significant overall effect for the ego-depletion effect pro-
vides confirmatory evidence for the acute decrements in self-
control task performance observed in experiments adopting the
strength model and the dual-task paradigm. Moderator analyses
examining the effect for both depleting and dependent tasks in
different spheres of self-control also supported the presence of the
effect. As predicted, any variation in the effect across spheres was
due to differences in the magnitude rather than whether the effect
was present or absent. Furthermore, the effect did not vary across
studies in which the depleting or dependent tasks were from
identical or different spheres of self-control, providing further
evidence to support the generality of the effect. Findings are
consistent with a resource depletion model, which predicts that
performance decrements carry across different spheres of self-
control. It also corroborates the view that self-control draws from
a single, global resource and depletion is not an artifact of specific
spheres or tasks. The present synthesis therefore contributes to
knowledge by demonstrating that the ego-depletion effect exists,
its associated confidence intervals do not include trivial values,
and it is generalizable across spheres of self-control.
The analysis of sphere of task as a moderator provided impor-
tant information on whether certain spheres of ego-depleting task
were more demanding of self-control resources than others. Intro-
ducing sphere as a moderator was effective in resolving some of
the heterogeneity in the overall ego-depletion effect in some mod-
erator groups. The majority of the spheres exhibited homogeneous
ego-depletion effects and did not differ significantly from each
other. The general pattern of these results suggests that most tasks
are equally taxing of self-control resources. However, there were
some variations. Studies that adopted depleting tasks in the con-
trolling impulses sphere and dependent tasks in the cognitive
sphere exhibited smaller ego-depletion effect sizes than those in a
minority of other spheres. These moderator groups also exhibited
high levels of heterogeneity. In some cases, the heterogeneity
associated with the effects in these spheres was resolved when
moderator groups were formed from individual, frequently used
depleting (e.g., crossing out letters) and dependent (e.g., modified
Stroop, food taste test) controlling-impulses tasks. Some individ-
ual tasks also displayed significant heterogeneity and lower ego-
depletion effects (e.g., modified Stroop as a depleting task), but
excluding studies adopting this task in the moderator analyses for
task sphere (e.g., controlling impulses, affective) did not produce
homogeneous effects. The lower size and heterogeneity of the
effect in these spheres do not appear to be attributable to one
particular task.
Studies adopting choice and volition tasks as a dependent vari-
able exhibited much lower levels of depletion than studies using
tasks in other specific spheres of self-control. A possible reason for
this variation is that making choices may not place as many
demands on depleted individuals’ self-control resources as do
tasks in other spheres, but this does not appear to be the case when
such tasks are used as depleting tasks. On closer inspection, the
effect was based on a small sample, with the majority of effects
derived from studies using a particular type of consumer-choice
task (Masicampo & Baumeister, 2008; Pocheptsova, Amir, Dhar,
& Baumeister, 2009). These tasks focus on ego-depleted individ-
uals’ failure to select a compromise option in a buying scenario,
likely due to the use of simple rather than deep-level processing.
The strength of the effect may be reduced for these tasks, because
a substantial proportion of people are still able make correct
decisions even when depleted. This individual task may therefore
be one that is less effective in measuring self-control resource
depletion, but it appears to be an exception rather than the rule.
An inconsistency in the ego-depletion literature is whether the
ego-depletion effect is exclusively the result of engaging in tasks
that require suppressing impulses or overriding dominant re-
sponses or whether difficult or challenging tasks also have the
propensity to evoke ego depletion. We hypothesized that difficult
tasks vary in complexity and, therefore, the extent to which they
deplete self-control resources. Although classifying such tasks
according to their complexity proved challenging due to their
diversity, cognitive processing tasks that varied in the number of
processes required provided an opportunity to test the effect of task
complexity on ego depletion. Complex tasks required a greater
number of cognitive processes such as mixed-arithmetic opera-
tions or memory updating than did tasks such as rote memory
(Schmeichel, 2007), which required relatively few. Consistent with
expectations, dependent tasks classified as complex led to a larger
ego-depletion effect than tasks classified as simple. Although the
difference was borderline significant, these results provide prelim-
inary evidence that the degree of ego depletion evoked by cogni-
tive processing tasks is dependent on task complexity. Findings
indicate that the depletion of self-control resources is not exclu-
sively confined to tasks that require overriding impulses or habit-
ual responses but also applies to difficult or challenging tasks that
demand complex cognitive processing.
It was also important to test whether the type of control condi-
tion task with which the depleting self-control task was compared
affected the ego-depletion effect. We tested this by including the
type of control condition task used in dual-task paradigm experi-
ments as a moderator variable. Most experiments used a modified,
easier version of the depleting self-control task for the control
condition task (e.g., a congruent Stroop task or eating a tempting
food in an ostensible taste test). A common feature of these tasks
is that individuals are required to act in accordance with, rather
than override, habitual or dominant responses. However, some
experiments used individual differences and alternative tasks (e.g.,
math problems) as the control condition. The average ego-
depletion effect in the present meta-analysis did not differ with the
type of control condition adopted. Although the selection of an
appropriate control condition is an important feature of the dual-
task paradigm in order to effectively induce ego depletion, the
extent of resource depletion does not depend on the types of
control task identified in this analysis. The implication of these
findings is that there is some flexibility within the dual-task
paradigm in terms of the means to deplete resources as well as the
control conditions with which the depleting task is compared.
However, researchers must exercise caution when selecting tasks
so that the features that define them as depleting or nondepleting
are clearly identifiable and can be justified theoretically. Our
classification of tasks based on complexity of the processing
required may provide a useful guide in this regard.
Another feature of depleting self-control tasks likely to affect
ego depletion is the length of time spent on the task. It was
proposed, consistent with the conceptualization of self-control as a
limited resource, that spending longer on the initial self-control
task in the dual-task paradigm would consume more of the re-
source. Our meta-analysis revealed a marginally significant rela-
tionship between duration of depleting task and ego depletion in
the hypothesized direction. The small effect size indicates that task
duration accounted for relatively little variance in the ego-
depletion effect. Task duration should therefore be a minor con-
sideration when designing and evaluating ego-depletion experi-
ments. It is important to note that the task duration range of the
studies used in the present analysis was relatively narrow; many of
the tasks were relatively brief in duration. This is consistent with
the heavy focus on short-term self-control failure in experimental
tests of ego depletion. The relatively brief task duration means that
it was unlikely there would be an observable deterioration in
performance on the initial self-control task itself in these studies.
The identification of a decline in performance on the initial task
would be a useful additional index of resource depletion and would
be consistent with the long-term task performance decrements
observed in the mental fatigue and vigilance literature (Parasura-
man, 1979; See et al., 1995). Furthermore, no study has manipu-
lated initial task duration in a dual-task paradigm and examined its
effects on performance of both initial and second tasks. The
present analysis provides the impetus for future research to exam-
ine extended task duration as a moderator of ego depletion.
We conducted analyses to test relations between significant
moderator variables in order to provide a critical evaluation of
potential confounding effects due to interrelationships among
moderators. In general, our analysis revealed few significant rela-
tions providing evidence that the moderators identified were
unique and the analyses were not confounded. We found two
instances in which a relationship between moderators had a de-
monstrable effect on the moderator analyses. We found that ex-
periments that did not report an interim period between tasks
exhibited weaker ego-depletion effects than those that reported an
interim period or the completion of questionnaires between tasks.
This finding was contrary to the recovery hypothesis, which sug-
gests that an interim period between tasks would permit partial
recovery of self-control resources and lead to better performance
on the second task. Our analysis of relations between moderators
suggested that this effect may have been confounded by a signif-
icant relationship between the interim period and task sphere
moderators. The analysis revealed that the inclusion of studies
adopting depleting tasks in the controlling impulses sphere that did
not report an interim period may have biased the ego-depletion
effect size downward. Reasons why researchers controlling im-
pulses depleting tasks were less likely to report an interim period
are unclear and cannot be ascertained from the current analysis.
However, there was evidence that this did not account for all the
variation in the ego-depletion effect across the interim period. An
important consideration when interpreting these findings is that
our coding of the “no interim period reported” category for this
moderator was relatively crude. For example, it was possible that
studies coded as reporting no interim period may have included an
interim period or filler task between tasks in their experiment but
the authors failed to mention doing so. Full reporting of the
duration and experimental requirements (e.g., filler tasks, ques-
tionnaire completion) between tasks is advocated to provide more
robust data for evaluating whether intertask period moderates the
ego-depletion effect.
We also found that presenting tasks in the dual-task paradigm by
the same or different experimenters moderated the ego-depletion
effect. Contrary to predictions and the experimenter-demand ac-
count, the ego-depletion effect size was larger among studies in
which the tasks were administered by different experimenters.
Analysis of relations between moderator variables suggested that a
significant association between the experiment presentation and
sphere of depleting task moderators may have confounded the
effect. We found that studies adopting social processing depleting
tasks were more likely to use different experimenters than studies
adopting tasks in other spheres. Our analysis revealed that inclu-
sion of studies adopting social processing depleting tasks tended to
bias the ego-depletion effect size for different experimenters up-
ward. This indicated that the association between these moderators
was a result of a methodological artifact of particular studies rather
than a systematic variation between studies. However, it must be
stressed that the pattern of the differences was identical for the task
presentation moderator analyses within the social processing and
other tasks moderator groups. This means that the inclusion of
these particular tasks was not exclusively responsible for the
significantly larger ego-depletion effect among studies using dif-
ferent experimenters. Furthermore, the analyses including different
experimenters comprised comparatively small samples of studies
and should be interpreted with caution. On the basis of this
analysis, researchers adopting the dual-task paradigm should in-
clude additional checks to ascertain the extent of participants’
perceived fulfillment of experimenter demands. Although partici-
pants in the present sample of studies were typically probed for
suspicion regarding links between the two tasks, few studies re-
ported whether participants thought they had satisfied the experi-
menter’s demands after the initial task. In addition, few studies
reported effort exerted on both the initial and second tasks. This
would provide useful information to complement the use of dif-
ferent experimenters as a means to evaluate the experimenter-
demand hypothesis.
Alternative Explanations
Our analysis also examined motivation, fatigue, self-efficacy,
and affect as alternative explanations for the ego-depletion effect
and how these explanations compared with the limited-resource
account offered by the strength model. Present findings provide
clear evidence that the impairment of task performance observed
in ego-depletion experiments coincides with increased perceptions
of fatigue, difficulty, and effort. Such perceptions substantiate the
greater demand that self-control tasks relative to control condition
tasks place on individuals’ resources and are often used as manip-
ulation checks in the dual-task paradigm (Baumeister et al., 1998;
Govorun & Payne, 2006). These findings also corroborate research
in related fields. For example, significant relations have been
found between cognitive performance and other indices of fatigue
like sleep deprivation (Barber, Munz, Bagsby, & Powell, 2009;
Drummond et al., 2005) and mental fatigue (Ackerman & Kanter,
2009). Fatigue therefore serves as an indicator of the increased
demands that self-control tasks place on individuals. This is not
inconsistent with a limited resource account for self-control fail-
ure. Self-regulation requires considerable effort, placing increased
demands on self-control resources. The depletion of resources is
likely to lead to subjective fatigue and a reason why subsequent
attempts at self-control fail. The prospect of future tasks will also
be perceived as more daunting when the individual is fatigued, so
future self-control tasks appear more difficult. Fatigue is expected
to serve as a mediating factor in the effect of self-control resource
depletion on task performance (Muraven et al., 1998), but there
have been no formal tests of this effect.
Reduced motivation has been proposed as a further alternative
explanation for self-regulatory failure, and fatigue may play an
important supporting role. According to a motivation-only expla-
nation, performing difficult and effortful tasks leads to a state of
mental fatigue and reduces the perceived importance of subsequent
task goals relative to the expected effort required. This results in
decreased motivation to perform subsequent tasks that are per-
ceived to be difficult, effortful, and fatiguing. By analogy, provid-
ing incentives or emphasizing the importance of task goals in-
creases motivation to engage in such tasks by providing reasons to
overcome fatigue and invest effort in attaining task goals. Consis-
tent with this theory, present findings indicate that incentives or
raising the importance of task goals reduced the deleterious effect
of ego depletion on self-control task performance. Motivation
therefore provides a viable explanation for the decrements in task
performance observed in ego-depletion experiments.
According to the strength model, the limited resource and mo-
tivational explanations for ego depletion are not irreconcilable.
The depletion of a limited self-control resource may be involved in
inducing perceptions of fatigue, and this in turn results in de-
creased motivation to exert future self-control and regulatory fail-
ure on subsequent tasks. There are two reasons for this effect from
the strength model perspective. First, depletion of self-control
resources leads to comparatively higher perceptions of the effort
required to engage in future acts of self-control, because doing so
is relatively costly when resources are scarce compared to when
they are plentiful. Second, engaging in self-control tasks reduces
the availability of resources. This leads to increased motivation to
conserve resources for times of need and reduced willingness to
allocate further resources on tasks perceived to be unimportant. In
contrast, providing incentives or increasing the importance of a
task may result in increased motivation to commit further self-
control resources to reach task goals in spite of the relatively high
cost. Individuals are therefore more inclined to allocate their
precious reserves because the task goal is perceived to be well
worth the effort. However, while this integrated motivation and
limited resource account may offer a solution in drawing together
these competing explanations, the present synthesis does not pro-
vide unequivocal support for this model. The evidence presented
here suggests that a motivation-only account of self-regulatory
failure in the dual-task paradigm holds as a viable alternative
explanation to the strength model.
The problem is that self-control capacity as a limited resource
has only been measured indirectly, meaning that the ego-depletion
effect has been exclusively inferred from postdepletion perfor-
mance on self-control tasks. Current data, therefore, cannot verify
whether the motivation-only and strength models provide exclu-
sive explanations for the performance decrements observed on
self-control tasks in ego-depletion experiments. Both models sug-
gest that engaging in debilitating tasks is a source of fatigue that
results in reduced motivation to engage in subsequent tasks. The
motivation-only account suggests that the fatigue leads to a per-
ceived imbalance between the effort required to achieve the goal
and its value. The strength model suggests that this imbalance is
due to reduced capacity to allocate depleted self-control resources
to the task and increased motivation to conserve those resources.
Until an objective measure of resource depletion has been identi-
fied, evidence to reconcile these competing explanations for the
ego-depletion effect is likely to remain elusive.
Self-efficacy has also been proposed as an alternative explana-
tion for ego depletion. People may have sufficient outcome ex-
pectancies regarding self-control tasks, but their perceived capac-
ity to exert the necessary effort to attain task goals is diminished
when they are in an ego-depleted state. The present analysis did
not find a significant effect of self-efficacy on ego depletion.
Furthermore, one study that attempted to change self-efficacy
beliefs by providing bogus feedback regarding task performance
found no significant effects (H. M. Wallace & Baumeister, 2002).
Although the limited number of tests does not unequivocally rule
out self-efficacy as a potential mediator of ego depletion, there is
little support for it based on the evidence presented in the literature
and in the current analysis.
A further alternative explanation for a decrease in performance
across tasks may be due to negative affect. Present results revealed
that negative affect is significantly related to ego depletion, with a
small, homogeneous effect size. It seems that engaging in self-
control tasks may induce negative affect, regardless as to whether
the task itself requires affect regulation, which is consistent with
expectations given their aversive nature (Tice & Bratslavsky,
2000). This would point to an explanation in which ego depletion
was the result of coping with a negative affective state. Recent
research has suggested that process behind the link between neg-
ative affective states and self-control failure is due to the depleting
nature of active mood regulation. People are not able to commit
resources to the second self-control task because attempts to reg-
ulate the negative affect induced by the initial task drain self-
control resources (Bruyneel et al., 2009). Few studies in the
present sample, however, reported relations between negative af-
fect and task performance. This negated a formal test of mediation
of the effect of ego depletion on task performance by negative
affect. Although there are tests of this mediation effect in the
ego-depletion literature (e.g., Gailliot et al., 2006), they are rela-
tively uncommon, and researchers are encouraged to investigate
the mediating role of negative affect in future ego-depletion stud-
Engaging in effortful, aversive tasks in ego-depletion experi-
ments was not related to positive affect. Ego depletion does not
undermine positive mood in the same way that it appears to induce
negative affect. However, research has found that inducing positive
affect is effective in overcoming the deleterious effects of self-control
resource depletion (Tice, Baumeister, Shmueli, & Muraven, 2007).
Positive affect may lead to renewed vigor toward tasks such that
individuals increase their efforts and motivation to expend self-control
resources in their pursuit. Alternatively, the induction of a positive
mood may obviate the need to commit self-control resources to
overcome the negative affect induced by the self-control tasks. In
contrast, there is evidence that inducing a positive mood does not
promote better self-control (Schmeichel & Vohs, 2009). Tests of
positive affect induction as means to overcome ego depletion are
relatively scarce, and additional empirical investigation is war-
ranted to resolve the inconsistency.
Additional Strength-Model Hypotheses
The present analysis tested additional hypotheses that have
emerged from the strength model. These hypotheses are meant to
provide a more complete explanation of self-regulatory failure
from a resource depletion perspective and assist in developing an
understanding of the mechanisms that underpin the ego-depletion
effect. In this section we evaluate the extent to which these
hypotheses are supported by the present meta-analysis and the
implications for the strength model.
Conservation. The strength model proposes that people at-
tempt to conserve self-control resources when they are aware of
forthcoming demands on their reserves. The depletion of self-
control in the strength model is viewed as temporary and may
result from complete depletion of finite self-control reserves or,
alternatively, partial depletion when people conserve their reserves
for future exertions (Baumeister & Heatherton, 1996; Muraven &
Baumeister, 2000). Present results support the partial-depletion
model. Individuals tend to modulate the application of self-control
resources when faced with the prospect of future demand. This
tendency is clearly adaptive from a limited resource perspective, as
it maximizes the application of self-control resources and produces
economy of effort. It is also consistent with an integrated motiva-
tional and limited resource account for ego depletion. Just as
incentives can increase motivation to invest increased effort and
allocate further self-control reserves postdepletion, future expec-
tation motivates the conservation of resources, particularly when
those reserves are depleted. This is because, when one’s reserves
are depleted, even a small outlay of resources is considered rela-
tively costly. The sample of the effect sizes testing this hypothesis
is comparatively small and comes from two studies (Muraven,
Shmueli, & Burkely, 2006; Tyler & Burns, 2009), so it is impor-
tant that this effect is replicated in a more diverse range of
self-control tasks.
Training. According to the strength model, regular training
on tasks requiring self-control will improve a person’s capacity to
exert self-control, just as training a muscle increases its strength
and endurance (Baumeister & Heatherton, 1996; Baumeister et al.,
2000). As expected, our analysis indicated that regular training
resulted in significantly superior postdepletion self-control perfor-
mance relative to that of untrained controls. A diverse range of
training techniques to improve self-control was identified. This is
important, as it not only supports the training hypothesis but
provides further support for the premise that self-control draws
from a common, global resource. Just as tasks from different
spheres can deplete self-control strength, so regular training on
self-control tasks in different spheres can help increase the capac-
ity to exert self-control.
The mechanism that underlies the training hypothesis remains
elusive. As yet, it is unclear whether training improves efficiency
in the application of finite self-control resources to self-control
tasks, perhaps through the development of self-control “skills,” or
whether the pool is just extended such that more of the self-control
resource is available. Future tests may examine the effect of
training on repeated self-control tasks that may address the effi-
ciency versus “extended pool” question. This may also shed light
on whether self-control is akin to a skill rather than a limited
resource. Research may also examine whether ceasing training
results in self-control capacity returning to baseline. This will
provide useful further information on the processes by which
training affects ego depletion and the longevity of the effect.
Finally, future studies should seek to investigate whether the increase
in self-control capacity as a result of training is linearly proportional
to the duration of the training period. Preliminary evidence has been
provided by Oaten and Cheng (2006b, 2007), who found steadily
decreasing levels of ego depletion within trained participants at mul-
tiple time points during an extended training intervention. Additional
multicohort within-participants training studies may provide further
data to support this hypothesis.
Recovery. In the original conceptualization of the strength
model, Baumeister and colleagues (Baumeister & Heatherton,
1996; Muraven & Baumeister, 2000) proposed that sufficient
recovery is necessary after resource depletion for self-regulatory
capacity to be restored. We investigated this in present analysis by
coding studies as those that required the completion of question-
naires between tasks, equivalent to a brief rest, a filler task, or rest
period between self-control tasks, and those that reported no in-
terim period. Ego depletion did not vary across the questionnaire
completion and filler task groups; however, contrary to hypothe-
ses, the no-reported-interim group had lower levels of ego deple-
tion. These results seem to suggest that immediate engagement in
tasks results in less resource depletion, which is inconsistent with
the recovery hypothesis. As we emphasized earlier, our coding of
intertask duration was relatively crude, as there is likely to be
considerable interindividual variation in questionnaire and filler
task completion time. Furthermore, the nonreporting of an interim
period is not equivalent to the absence of an interim period. A
more effective means of testing the role of recovery on ego
depletion would be the inclusion of recovery or relaxation manip-
ulations in the dual-task paradigm. Studies have demonstrated
reduced ego depletion among participants receiving recovery ma-
nipulations (Oaten et al., 2008; Tyler & Burns, 2008), but there
were insufficient effects for a meta-analysis. There is clearly a
need for additional studies testing the role of recovery and relax-
ation. There is also a need to examine the effect of chronic lack of
recovery on self-control capacity. For example, Baumeister and
Heatherton (1996) suggested that sleep may assist in recovery
from self-control resource depletion and that sleep deprivation
may lead to failure to recover and a chronic state of ego depletion
(C. Anderson, 2009; Barber et al., 2009; Wright, 2009).
Glucose and glucose supplementation. A recent avenue of
inquiry in the pursuit of the mechanisms that govern ego depletion
has been the identification of physiological mediators and moder-
ators. Most prominent is the proposed role of blood glucose as a
mediator of the effect of ego depletion on self-control task perfor-
mance in the dual-task paradigm (Dvorak & Simons, 2009). Gail-
liot and Baumeister (2007a) have suggested that as self-control
tasks require increased cerebral functioning, they may cause a
concomitant rise in the demand for glucose in the brain (Benton,
Parker, & Donohoe, 1996; Green, Elliman, & Rogers, 1997). They
proposed that changes in blood glucose levels as a consequence of
engaging in self-control tasks in dual-task paradigm experiments
may serve as a proxy measure of this demand. Results of the
present meta-analysis indicate that blood glucose levels were sig-
nificantly associated with ego depletion. However, there are two
caveats to bear in mind when interpreting these findings. First, the
data are from a limited number of effect sizes (k5) and only two
studies (Dvorak & Simons, 2009; Gailliot, Baumeister, et al.,
2007), albeit from separate laboratories. Second, the studies did
not measure glucose consumption in the brain, and this would be
necessary to validate the proposed mechanism. Conclusions based
on these data should not, therefore, be treated as unequivocally
supportive of glucose as a physiological indicator of the ego-
depletion effect.
An additional finding in the present analysis was that glucose
supplementation allayed the effects of ego depletion on self-
control task performance. This also provides limited evidence for
glucose as a control mechanism for self-regulatory resource de-
pletion. Nevertheless, there may be a number of plausible alterna-
tive explanations for this effect. For example, the consumption of
beverages rich in sugar is associated with increased positive affect
(Benton, 2002) and may have resulted in a concomitant increase in
motivation to perform subsequent tasks. Similarly, researchers
have proposed that it is the perception of glucose in the mouth
cavity, as opposed to ingestion and substrate use, that leads to
increased exertion on tasks (Pottier, Bouckaert, Gilis, Roels, &
Derave, 2010). In summary, although present results provide pre-
liminary evidence that blood glucose and glucose supplementation
are implicated in the effect of ego-depletion on self-control task
performance, the aforementioned caveats and alternative explana-
tions mean that these findings should be regarded as tentative and
not indicative of direct evidence for role of glucose as a mecha-
nism for the ego-depletion effect.
Limitations and Future Directions
A limitation of the present analysis is that some of the effect
sizes reported were based on relatively small sample sizes. As
simulation studies have demonstrated that meta-analyses based on
small samples of studies tend to bias the effect size upward
(Reynolds & Day, 1984), caution should be exercised in interpret-
ing the size of the effects in some of the subanalyses in the present
research. In particular, analyses of the conservation, training, and
glucose supplementation moderator groups were based on sets of
studies numbering as few as five. Although these effects were
corrected for statistical artifacts, the potential for one or two effects
to alter the size and distribution of the averaged effect size re-
mains. Replication of these effects in different spheres of self-
control should be a priority for future investigation.
The present meta-analysis has assisted in identifying gaps in the
ego-depletion literature and priorities for future research. An out-
standing issue is the longevity of the ego-depletion effect and the
need to reconcile the long- and short-term effects of self-control
resource depletion. Baumeister and coworkers (Baumeister et al.,
1998; Muraven et al., 1998) explicitly align the strength model
with short-term depletion of self-control resources. It is clear that
the dual-task paradigm and experimental tests of the theory have
been geared toward testing such a model by adopting tasks typi-
cally less than 10 minutes in duration (Vohs et al., 2008). How-
ever, it is unclear what effects chronic self-control efforts have on
ego depletion. It is difficult to distinguish between conditions that
lead to the observation of a training effect and improvements in
self-control and those that lead to long-term fatigue and decre-
ments in self-control performance. For example, Vohs and Heath-
erton (2000) proposed that “repeated attempts at inhibition render
a person especially vulnerable to situational temptations” (p. 249).
They surmised that long-term dieters were compelled to engage in
the chronic inhibition of the impulse to eat and would therefore be
more vulnerable to depletion. In contrast, the training hypothesis
proposes that long-term practice on self-control increases resis-
tance to the deleterious effects of self-control exertion on task
performance (Muraven et al., 1999; Oaten & Cheng, 2006a, 2006b,
2007). If the latter is the case, would dieters not be expected to be
more effective in controlling their desire to eat if they have had
long-term experience with resisting temptation? A possible reason
for this inconsistency may be differences in the relative frequency
and success of the long-term inhibitory experiences. Training
effects have usually been observed after a series of relatively
discrete and planned acts of self-control with sufficient recovery
allowing for gradual improvement. Furthermore, in training stud-
ies participants are invariably successful in the self-control tasks
they perform. The self-control acts are therefore experienced as
finite quanta of successful self-regulation with sufficient time for
recovery before the next act. In contrast, long-term acts of self-
control in real-world contexts such as dieting are likely lead to
frequent and unplanned attempts to resist temptations throughout
the day. Furthermore, there is likely to be greater variation in the
frequency of success, and dieters may experience long-term failed
attempts to self-regulate with little opportunity to recover. In
strength model vernacular, this may be the equivalent of “over-
training” resulting in a chronically-depleted state. Empirical sup-
port for these proposed mechanisms is necessary to resolve this
Further research should also identify the conditions that lead to
the attenuating effect of motivation on ego depletion. If motivation
compels individuals to tap further into their finite self-control
reserves, as suggested by the findings of the present analysis, it is
unlikely that they will be able to stave off the depleting effects of
self-control tasks indefinitely. Evidence for this comes from re-
search demonstrating that performance on vigilance tasks declines
steadily over time (See et al., 1995). The effect of motivation on
repeated bouts of self-control exertion needs to be tested to delin-
eate the boundary conditions of motivation as a strategy to allay
ego depletion. It is also important that the conservation of self-
control resources is studied in conjunction with motivation. For
example, will people anticipating a future self-control task still
conserve their self-control resources if they are provided with an
incentive to perform the second task? Or will they forgo the
tendency to conserve resources and expend more of their reserves
because they have increased motivation to perform the task?
The present analysis did not include individual differences in
self-control as a moderator of the ego-depletion effect due to the
limited number of tests of the effect. Research suggests that
dispositional self-control may moderate the deleterious effects of
situationally induced self-control resource depletion on subsequent
task performance (Dvorak & Simons, 2009; Friese & Hofmann,
2009; Gailliot & Baumeister, 2007b; Gailliot, Schmeichel, &
Maner, 2007). This is consistent with a limited capacity model of
self-control. Dispositional self-control likely reflects the extent of
an individual’s self-control reserves and therefore determines the
potential resources available for allocation to self-control tasks.
This trait-level capacity has also been related to long-term self-
regulatory efforts (Tangney et al., 2004). Future research should
provide further tests of the interaction between ego depletion and
trait self-control on task performance using the dual-task para-
digm. This will assist in further integrating dispositional and
situational influences on self-control performance.
The strength model may provide only a partial explanation for
self-control failure. Current data support a limited resource ac-
count for short-term self-control depletion but do not unequivo-
cally support or falsify alternative explanations such as
motivation-only and coping with negative affect. Integrating the
strength model with other theories may provide more comprehen-
sive explanations of self-regulation. For example, theorists have
recently proposed the integration of the strength model and dual-
process models of behavioral decision making (Dvorak & Simons,
2009; Hagger, Wood, Stiff, & Chatzisarantis, 2009; Masicampo &
Baumeister, 2008; Pocheptsova et al., 2009). In such models,
behavior is viewed as a function of effortful, intentional decisions
that require deliberation and information processing (reflective
route) and reactive responses that are spontaneous, unplanned and
reliant on heuristic processing (impulsive route; Hofmann, Friese,
& Strack, 2009; Hofmann, Friese, & Wiers, 2008; Strack &
Deutsch, 2004). Self-control is conceptualized as a deliberate and
intentional attempt to override and gain control over impulsive
responses driven by situational cues and immediate rewards. En-
gaging in behaviors that require overcoming dominant responses
and focused commitment on long-term goal attainment is therefore
most effectively controlled by the reflective route requiring moti-
vation and effort. Resource depletion may be a mechanism that
determines whether actions are determined by reflective rather
than impulsive routes to behavior. A lack of self-control resources
decreases the propensity of the reflective pathway to override the
influence of the impulsive pathway on behavior. This is likely to
result in behavior becoming increasingly determined by the im-
pulsive route and in reduced persistence on long-term, planned
behaviors (Dvorak & Simons, 2009). These developments are the
beginnings of endeavors to unify theories of self-control to resolve
inconsistencies and provide complementary explanations of be-
The present meta-analysis provides evidence to suggest that the
strength model is a useful explanatory system with which to
understand self-control, but further refinements may be necessary,
particularly when it comes to the identification of mechanisms. For
example, the present analysis could not unequivocally rule out
other explanations for ego depletion, such as motivation only and
negative affect. These alternatives are not necessarily inconsistent
with a resource depletion account and may provide insight into the
possible processes that underpin ego depletion. It is also important
to note that support for self-control as a limited resource has only
been inferred indirectly from performance decrements on self-
control tasks (Schmeichel, Demaree, Robinson, & Pu, 2006). Iden-
tifying physiological analogs for ego depletion may offer a poten-
tial solution. The finding that blood glucose varies with ego
depletion, for example, provides preliminary evidence linking self-
control strength to a physical resource (Dvorak & Simons, 2009;
Gailliot, Baumeister, et al., 2007). These results have been com-
plemented by research that has linked physiological analogs for
effort including electromyographic activity (Bray et al., 2008),
cardiovascular response (Segerstrom & Nes, 2007; Wright et al.,
2008), and galvanic skin response (Sheppes, Catran, & Meiran,
2009) with ego depletion. Such evidence is consistent with
Baumeister et al.’s (1998) prediction that it is “implausible that ego
depletion would have no physiological aspect or correlates at all”
(p. 1263). In practical terms, these explanations and mechanisms
may provide insight into strategies that can be used to overcome
ego depletion. Intervention techniques to increase motivation and
promote regular practice on self-control tasks are means that have
been suggested to improve self-control capacity (Hagger et al.,
2009; Hui et al., 2009; Muraven & Slessareva, 2003). The strength
model offers promise in identifying strategies to minimize short-
term decrements in self-control and assist in developing interven-
tions that foster better self-regulation.
For a complete list of list of studies included in the meta-analysis, go to
Ackerman, P. L., & Kanter, R. (2009). Test length and cognitive fatigue:
An empirical examination of effects on performance and test-taker
reactions. Journal of Experimental Psychology: Applied, 15, 163–181.
Aguinis, H., & Pierce, C. A. (1998). Testing moderator variable hypotheses
meta-analytically. Journal of Management, 24, 577–592. doi:10.1177/
Ainslie, G. (1975). Specious reward: A behavioral theory of impulsiveness
and impulse control. Psychological Bulletin, 82, 463–496. doi:10.1037/
Ajzen, I. (1985). From intentions to actions: A theory of planned behavior.
In J. Kuhl & J. Beckmann (Eds.), Action-control: From cognition to
behavior (pp. 11–39). Heidelberg, Germany: Springer-Verlag.
Anderson, C. (2009). The impact of sleep on dealing with daily stressors:
A need for controlled laboratory evidence. Stress and Health. Advance
online publication. doi:10.1002/smi.1301
Anderson, J. R. (1982). Acquisition of cognitive skill. Psychological Re-
view, 89, 369406. doi:10.1037/0033-295X.89.4.369
Bagozzi, R. P. (1992). The self-regulation of attitudes, intentions and
behavior. Social Psychology Quarterly, 55, 178–204.
Barber, L., Munz, D., Bagsby, P., & Powell, E. (2009). Sleep consistency
and sufficiency: Are both necessary for less psychological strain? Stress
and Health. Advance online publication. doi:10.1002/smi.1292
Barnett, V., & Lewis, T. (1994). Outliers in statistical data (3rd ed.).
Chichester, England: Wiley.
Baumeister, R. F. (2002). Ego depletion and self-control failure: An energy
model and the self’s executive function. Self and Identity, 1, 129–136.
Baumeister, R. F. (2003). Ego depletion and self-regulation failure: A
resource model of self-control. Alcoholism: Clinical and Experimental
Research, 27, 281–284. doi:10.1097/01.alc.0000060879.61384.a4
Baumeister, R. F. (2005). The cultural animal: Human nature, meaning,
and social life. New York, NY: Oxford University Press.
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, 1252–1265. doi:10.1037/0022-3514.74.5.1252
Baumeister, R. F., Gailliot, M. T., DeWall, C. N., & Oaten, M. (2006).
Self-regulation and personality: How interventions increase regulatory suc-
cess, and how depletion moderates the effects of traits on behavior. Journal
of Personality, 74, 1773–1801. doi:10.1111/j.1467-6494.2006.00428.x
Baumeister, R. F., & Heatherton, T. F. (1996). Self-regulation failure:
An overview. Psychological Inquiry, 7, 1–15. doi:10.1207/
Baumeister, R. F., Heatherton, T. F., & Tice, D. M. (1994). Losing control:
How and why people fail at self-regulation. San Diego, CA: Academic.
Baumeister, R. F., Muraven, M., & Tice, D. M. (2000). Ego depletion: A
resource model of volition, self-regulation, and controlled processing.
Social Cognition, 18, 130–150.
Baumeister, R. F., Sparks, E. A., Stillman, T. F., & Vohs, K. D. (2008). Free
will in consumer behavior: Self-control, ego depletion, and choice. Journal
of Consumer Psychology, 18, 4–13. doi:10.1016/j.jcps.2007.10.002
Baumeister, R. F., & Vohs, K. D. (2007). Self-regulation, ego-depletion,
and motivation. Social and Personality Psychology Compass, 1, 115–
128. doi:10.1111/j.1751-9004.2007.00001.x
Baumeister, R. F., Vohs, K. D., & Tice, D. M. (2007). The strength model
of self-control. Current Directions in Psychological Science, 16, 351–
355. doi:10.1111/j.1467-8721.2007.00534.x
Benton, D. (2002). Carbohydrate ingestion, blood glucose and mood.
Neuroscience and Biobehavioral Reviews, 26, 293–308. doi:10.1016/
Benton, D., Parker, P. Y., & Donohoe, R. T. (1996). The supply of glucose
to the brain and cognitive functioning. Journal of Biosocial Science, 28,
463–479. doi:10.1017/S0021932000022537
Boksem, M. A. S., Meijman, T. F., & Lorist, M. M. (2006). Mental fatigue,
motivation and action monitoring. Biological Psychology, 72, 123–132.
Bray, S. R., Ginis, K. A. M., Hicks, A. L., & Woodgate, J. (2008). Effects
of self-regulatory strength depletion on muscular performance and EMG
activation. Psychophysiology, 45, 337–343. doi:10.1111/j.1469-
Bruyneel, S. D., Dewitte, S., Franses, P. H., & Dekimpe, M. G. (2009). I
felt low and my purse feels light: Depleting mood regulation attempts
affect risk decision making. Journal of Behavioral Decision Making, 22,
153–170. doi:10.1002/bdm.619
Bruyneel, S. D., Dewitte, S., Vohs, K. D., & Warlop, L. (2006). Repeated
choosing increases susceptibility to affective product features. Interna-
tional Journal of Research in Marketing, 23, 215–225. doi:10.1016/
Burkley, E. (2008). The role of self-control in resistance to persuasion.
Personality and Social Psychology Bulletin, 34, 419431. doi:10.1177/
Cameron, C. (1973). A theory of fatigue. Ergonomics, 16, 633–648.
Carver, C. S., & Scheier, M. F. (1998). On the self-regulation of behavior.
New York, NY: Cambridge University Press.
Ciarocco, N. J., Sommer, K. L., & Baumeister, R. F. (2001). Ostracism and
ego depletion: The strains of silence. Personality and Social Psychology
Bulletin, 27, 1156–1163. doi:10.1177/0146167201279008
Cochran, W. G. (1952). The
test of goodness of fit. Annals of Mathe-
matical Statistics, 23, 315–345. doi:10.1214/aoms/1177692778
Cohen, J. (1987). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Erlbaum.
DeCoster, J. (2004). Meta-analysis notes. Retrieved from http://www.stat-
DeCoster, J., & Claypool, H. M. (2004). A meta-analysis of priming effects
on impression formation supporting a general model of informational
biases. Personality and Social Psychology Review, 8, 2–27. doi:10.1207/
DeWall, C. N., Baumeister, R. F., Gailliot, M. T., & Maner, J. K. (2008).
Depletion makes the heart grow less helpful: Helping as a function of
self-regulatory energy and genetic relatedness. Personality and Social
Psychology Bulletin, 34, 1653–1662. doi:10.1177/0146167208323981
DeWall, C. N., Baumeister, R. F., Stillman, T. F., & Gailliot, M. T. (2007).
Violence restrained: Effects of self-regulation and its depletion on ag-
gression. Journal of Experimental Social Psychology, 43, 62–76. doi:
Drummond, S. P., Bischoff-Grethe, A., Dinges, D. F., Ayalon, L., Mednick,
S. C., & Meloy, M. J. (2005). The neural basis of the psychomotor vigilance
task. Sleep, 28, 1059–1068.
Duckworth, A. L., & Seligman, M. E. P. (2005). Self-discipline outdoes IQ
in predicting academic performance of adolescents. Psychological Sci-
ence, 16, 939–944. doi:10.1111/j.1467-9280.2005.01641.x
Dvorak, R. D., & Simons, J. S. (2009). Moderation of resource depletion
in the self-control strength model: Differing effects of two modes of
self-control. Personality and Social Psychology Bulletin, 35, 572–583.
Eisenberg, N., Valiente, C., Fabes, R. A., Smith, C. L., Reiser, M.,
Shepard, S. A., . . . Cumberland, A. J. (2003). The relations of effortful
control and ego control to children’s resiliency and social functioning.
Developmental Psychology, 39, 761–776. doi:10.1037/0012-
Fennis, B. M., Janssen, L., & Vohs, K. D. (2009). Acts of benevolence: A
limited-resource account of compliance with charitable requests. Journal
of Consumer Research, 35, 906–924. doi:10.1086/593291
Field, A. P. (2003). The problems using fixed-effects models of meta-
analysis on real-world data. Understanding Statistics, 2, 105–124. doi:
Finkel, E. J., & Campbell, W. K. (2001). Self-control and accommodation
in close relationships: An interdependence analysis. Journal of Person-
ality and Social Psychology, 81, 263–277. doi:10.1037//0022-
Finkel, E. J., Dalton, A. N., Campbell, W. K., Brunell, A. B., Scarbeck,
S. J., & Chartrand, T. L. (2006). High-maintenance interaction: Ineffi-
cient social coordination impairs self-regulation. Journal of Personality
and Social Psychology, 91, 456475. doi:10.1037/0022-3514.91.3.456
Finkel, E. J., DeWall, C. N., Slotter, E. B., Oaten, M., & Foshee, V. A.
(2009). Self-regulatory failure and intimate partner violence perpetra-
tion. Journal of Personality and Social Psychology, 97, 483–499. doi:
Fischer, P., Greitemeyer, T., & Frey, D. (2007). Ego depletion and positive
illusions: Does the construction of positivity require regulatory re-
sources? Personality and Social Psychology Bulletin, 33, 1306–1321.
Fischer, P., Greitemeyer, T., & Frey, D. (2008). Self-regulation and selec-
tive exposure: The impact of depleted self-regulation resources on
confirmatory information processing. Journal of Personality and Social
Psychology, 94, 382–395. doi:10.1037/0022-3514.94.3.382
Fishbach, A., Friedman, R. S., & Kruglanski, A. W. (2003). Leading us not
unto temptation: Momentary allurements elicit overriding goal activa-
tion. Journal of Personality and Social Psychology, 84, 296–309. doi:
Fishbach, A., & Labroo, A. A. (2007). Be better or be merry: How mood
affects self-control. Journal of Personality and Social Psychology, 93,
158–173. doi:10.1037/0022-3514.93.2.158
Folkman, S., & Moskowitz, J. T. (2000). Positive affect and the other side
of coping. American Psychologist, 55, 647–654. doi:10.1037/0003-
Friese, M., & Hofmann, W. (2009). Control me or I will control you:
Impulses, trait self-control, and the guidance of behavior. Journal of
Research in Personality, 43, 795–805. doi:10.1016/j.jrp.2009.07.004
Friese, M., Hofmann, W., & Wanke, M. (2008). When impulses take over:
Moderated predictive validity of explicit and implicit attitude measures
in predicting food choice and consumption behaviour. British Journal of
Social Psychology, 47, 397–419. doi:10.1348/014466607x241540
Fujita, K., & Han, H. A. (2009). Moving beyond deliberative control of
impulses: The effect of construal levels on evaluative associations in
self-control conflicts. Psychological Science, 20, 799804. doi:10.1111/
Funder, D. C., Block, J. H., & Block, J. (1983). Delay of gratification:
Some longitudinal personality correlates. Journal of Personality and
Social Psychology, 44, 1198–1213. doi:10.1037/0022-3514.44.6.1198
Gailliot, M. T., & Baumeister, R. F. (2007a). The physiology of willpower:
Linking blood glucose to self-control. Personality and Social Psychol-
ogy Review, 11, 303–327. doi:10.1177/1088868307303030
Gailliot, M. T., & Baumeister, R. F. (2007b). Self-regulation and sexual
restraint: Dispositionally and temporarily poor self-regulatory abilities con-
tribute to failures at restraining sexual behavior. Personality and Social
Psychology Bulletin, 33, 173–186. doi:10.1177/0146167206293472
Gailliot, M. T., Baumeister, R. F., DeWall, C. N., Maner, J. K., Plant,
E. A., Tice, D. M., . . . Schmeichel, B. J. (2007). Self-control relies on
glucose as a limited energy source: Willpower is more than a metaphor.
Journal of Personality and Social Psychology, 92, 325–336. doi:
Gailliot, M. T., Peruche, M., Plant, E. A., & Baumeister, R. F. (2009).
Stereotypes and prejudice in the blood: Sucrose drinks reduce prejudice
and stereotyping. Journal of Experimental Social Psychology, 45, 288
290. doi:10.1016/j.jesp.2008.09.003
Gailliot, M. T., Plant, E. A., Butz, D. A., & Baumeister, R. F. (2007).
Increasing self-regulatory strength can reduce the depleting effect of
suppressing stereotypes. Personality and Social Psychology Bulletin, 33,
281–294. doi:10.1177/0146167206296101
Gailliot, M. T., Schmeichel, B. J., & Baumeister, R. F. (2006). Self-
regulatory processes defend against the threat of death: Effects of
self-control depletion and trait self-control on thoughts and fears of
dying. Journal of Personality and Social Psychology, 91, 4962. doi:
Gailliot, M. T., Schmeichel, B. J., & Maner, J. K. (2007). Differentiating
the effects of self-control and self-esteem on reactions to mortality
salience. Journal of Experimental Social Psychology, 43, 894–901.
Geeraert, N., & Yzerbyt, V. Y. (2007). How fatiguing is dispositional
suppression? Disentangling the effects of procedural rebound and ego-
depletion. European Journal of Social Psychology, 37, 216–230. doi:
Gordijn, E. H., Hindriks, I., Koomen, W., Dijksterhuis, A., & Van Knippen-
berg, A. (2004). Consequences of stereotype suppression and internal sup-
pression motivation: A self-regulation approach. Personality and Social
Psychology Bulletin, 30, 212–224. doi:10.1177/0146167203259935
Govorun, O., & Payne, B. K. (2006). Ego-depletion and prejudice: Sepa-
rating automatic and controlled components. Social Cognition, 24, 111–
136. doi:10.1521/soco.2006.24.2.111
Green, M. W., Elliman, N. A., & Rogers, P. J. (1997). The effects of food
deprivation and incentive motivation on blood glucose levels and cog-
nitive function. Psychopharmacology, 134, 88–94. doi:10.1007/
Grubbs, F. E. (1950). Sample criteria for testing outlying observations.
Journal of the American Statistical Association, 21, 27–58.
Hagger, M. S., Wood, C., Stiff, C., & Chatzisarantis, N. L. D. (2009). The
strength model of self-regulation failure and health-related behavior. Health
Psychology Review, 3, 208–238. doi:10.1080/17437190903414387
Hammer, E. Y. (2005). From the laboratory to the classroom and back: The
science of interpersonal relationships informs teaching. Journal of Social
and Clinical Psychology, 24, 3–10. doi:10.1521/jscp.
Hedges, L. V., & Olkin, I. (1983). Regression models in research synthesis.
American Statistician, 37, 137–140.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis.
Orlando, FL: Academic Press.
Henden, E. (2008). What is self-control? Philosophical Psychology, 21,
69–90. doi:10.1080/09515080701874092
Higgins, J. P. T., & Thompson, S. G. (2002). Quantifying heterogeneity in
a meta-analysis. Statistics in Medicine, 21, 1539–1558. doi:10.1002/
Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003).
Measuring inconsistency in meta-analyses. British Medical Journal,
327, 557–560. doi:10.1136/bmj.327.7414.557
Hofmann, W., Friese, M., & Strack, F. (2009). Impulse and self-control
from a dual-systems perspective. Perspectives on Psychological Science,
4, 162–176. doi:10.1111/j.1745-6924.2009.01116.x
Hofmann, W., Friese, M., & Wiers, R. W. (2008). Impulsive versus
reflective influences on health behavior: A theoretical framework and
empirical review. Health Psychology Review, 2, 111–137. doi:10.1080/
Hofmann, W., Rauch, W., & Gawronski, B. (2007). And deplete us not into
temptation: Automatic attitudes, dietary restraint, and self-regulatory
resources as determinants of eating behavior. Journal of Experimental
Social Psychology, 43, 497–504. doi:10.1016/j.jesp.2006.05.004
Huedo-Medina, T. B., Sa´nchez-Meca, J., Marı´n-Martı´nez, F., & Botella, J.
(2006). Assessing heterogeneity in meta-analysis: Qstatistic or I
Psychological Methods, 11, 193–206. doi:10.1037/1082-989X.11.2.193
Hui, S.-K. A., Wright, R. A., Stewart, C. C., Simmons, A., Eaton, B., &
Nolte, R. N. (2009). Performance, cardiovascular, and health behavior
effects of an inhibitory strength training intervention. Motivation and
Emotion, 33, 419434. doi:10.1007/s11031-009-9146-0
Hullett, C. R., & Levine, T. R. (2003). The overestimation of effect sizes
from Fvalues in meta-analysis: The cause and solution. Communication
Monographs, 70, 52–67. doi:10.1080/715114664
Hunter, J. E., & Schmidt, F. (1994). Methods of meta-analysis: Correcting
error and bias in research findings (2nd ed.). Newbury Park, CA: Sage.
Hunter, J. E., & Schmidt, F. L. (2000). Fixed effects vs. random effects
meta-analysis models: Implications for cumulative research knowledge
in psychology. International Journal of Selection and Assessment, 8,
275–292. doi:10.1111/1468-2389.00156
Johns, M., Inzlicht, M., & Schmader, T. (2008). Stereotype threat and
executive resource depletion: Examining the influence of emotion reg-
ulation. Journal of Experimental Psychology: General, 137, 691–705.
Joireman, J., Balliet, D., Sprott, D., Spangenberg, E., & Schultz, J. (2008).
Consideration of future consequences, ego-depletion, and self-control:
Support for distinguishing between CFC-Immediate and CFC-Future
sub-scales. Personality and Individual Differences, 45, 15–21. doi:
Kisamore, J. L., & Brannick, M. T. (2008). An illustration of the conse-
quences of meta-analysis model choice. Organizational Research Meth-
ods, 11, 35–53. doi:10.1177/1094428106287393
Koestner, R., Bernieri, F., & Zuckerman, M. (1992). Self-regulation and
consistency between attitudes, traits, and behaviors. Personality and
Social Psychology Bulletin, 18, 52–59. doi:0.1177/0146167292181008
Leith, K. P., & Baumeister, R. F. (1996). Why do bad moods increase
self-defeating behavior? Emotion, risk taking, and self-regulation. Jour-
nal of Personality and Social Psychology, 71, 1250–1267. doi:10.1037/
Levy, N. (2006). Addiction, autonomy and ego-depletion: A response to
Bennett Foddy and Julian Savulescu. Bioethics, 20, 16–20. doi:10.1111/
Lipsey, M. W. (2003). Those confounded moderators in meta-analysis:
Good, bad, and ugly. Annals of the American Academy of Political and
Social Science, 587, 6981. doi:10.1177/0002716202250791
Loewenstein, G. (1996). Out of control: Visceral influences on behavior.
Organizational Behavior and Human Decision Processes, 65, 272–292.
Logue, A. W. (1988). Research on self-control: An integrating frame-
work. Behavioral and Brain Sciences, 11, 665–709. doi:10.1017/
Lorist, M. M., Boksem, M. A. S., & Ridderinkhof, K. R. (2005). Impaired
cognitive control and reduced cingulate activity during mental fatigue.
Cognitive Brain Research, 24, 199–205. doi:10.1016/j.cogbrainres
Martijn, C., Alberts, H., Merckelbach, H., Havermans, R., Huijts, A., & De
Vries, N. K. (2007). Overcoming ego depletion: The influence of ex-
emplar priming on self-control performance. European Journal of Social
Psychology, 37, 231–238. doi:10.1002/ejsp.350
Masicampo, E. J., & Baumeister, R. F. (2008). Toward a physiology of
dual-process reasoning and judgment: Lemonade, willpower, and expen-
sive rule-based analysis. Psychological Science, 19, 255–260. doi:
Mayer, J. D., & Gaschke, Y. N. (1988). The experience and meta-
experience of mood. Journal of Personality and Social Psychology, 55,
102–111. doi:10.1037/0022-3514.55.1.102
Mead, N. L., Baumeister, R. F., Gino, F., Schweitzer, M. E., & Ariely, D.
(2009). Too tired to tell the truth: Self-control resource depletion and
dishonesty. Journal of Experimental Social Psychology, 45, 594–597.
Metcalfe, J., & Mischel, W. (1999). A hot/cool-system analysis of delay of
gratification: The dynamics of willpower. Psychological Review, 106,
3–19. doi:10.1037/0033-295X.106.1.3
Mischel, W. (1996). From good intentions to willpower. In P. M. Gollwit-
zer & J. A. Bargh (Eds.), The psychology of action: Linking cognition
and motivation to behavior (pp. 197–218). New York, NY: Guilford
Mischel, W., Shoda, Y., & Rodrieguez, M. L. (1989, May 26). Delay of
gratification in children. Science, 244, 933–938. doi:10.1126/
Moller, A. C., Deci, E. L., & Ryan, R. M. (2006). Choice and ego
depletion: The moderating role of autonomy. Personality and Social
Psychology Bulletin, 32, 1024–1036. doi:10.1177/0146167206288008
Muraven, M. (2008a). Autonomous self-control is less depleting. Journal
of Research in Personality, 42, 763–770. doi:10.1016/j.jrp.2007.08.002
Muraven, M. (2008b). Prejudice as self-control failure. Journal of Applied
Social Psychology, 38, 314–333. doi:10.1111/j.1559-1816.2007.00307.x
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of
limited resources: Does self-control resemble a muscle? Psychological
Bulletin, 126, 247–259. doi:10.1037/0033-2909.126.2.247
Muraven, M., Baumeister, R. F., & Tice, D. M. (1999). Longitudinal
improvement of self-regulation through practice: Building self-control
strength through repeated exercise. Journal of Social Psychology, 139,
446457. doi:10.1080/00224549909598404
Muraven, M., Collins, R. L., & Nienhaus, K. (2002). Self-control and
alcohol restraint: An initial application of the self-control strength
model. Psychology of Addictive Behaviors, 16, 113–120.
Muraven, M., Collins, R. L., Shiffman, S., & Paty, J. A. (2005). Daily
fluctuations in self-control demands and alcohol intake. Psychology of
Addictive Behaviors, 19, 140–147. doi:10.1037/0893-164x.19.2.140
Muraven, M., Gagne, M., & Rosman, H. (2008). Helpful self-control:
Autonomy support, vitality, and depletion. Journal of Experimental
Social Psychology, 44, 573–585. doi:10.1016/j.jesp.2007.10.008
Muraven, M., Pogarsky, G., & Shmueli, D. (2006). Self-control depletion
and the general theory of crime. Journal of Quantitative Criminology,
22, 263–277. doi:10.1007/s10940-006-9011-1
Muraven, M., Shmueli, D., & Burkley, E. (2006). Conserving self-control
strength. Journal of Personality and Social Psychology, 91, 524–537.
Muraven, M., & Slessareva, E. (2003). Mechanisms of self-control failure:
Motivation and limited resources. Personality and Social Psychology
Bulletin, 29, 894–906. doi:10.1177/0146167203253209
Muraven, M., Tice, D. M., & Baumeister, R. F. (1998). Self-control as a
limited resource: Regulatory depletion patterns. Journal of Personality
and Social Psychology, 74, 774–789. doi:10.1037/0022-3514.74.3.774
Neubach, B., & Schmidt, K. H. (2008). Main and interaction effects of
different self-control demands on indicators of job strain. Zeitschrift Fur
Arbeits-Und Organisationspsychologie, 52, 17–24. doi:10.1026/0932-
Oaten, M., & Cheng, K. (2006a). Improved self-control: The benefits of a
regular program of academic study. Basic and Applied Social Psychol-
ogy, 28, 1–16. doi:10.1207/s15324834basp2801_1
Oaten, M., & Cheng, K. (2006b). Longitudinal gains in self-regulation
from regular physical exercise. British Journal of Health Psychology,
11, 717–733. doi:10.1348/135910706X96481
Oaten, M., & Cheng, K. (2007). Improvements in self-control from finan-
cial monitoring. Journal of Economic Psychology, 28, 487–501. doi:
Oaten, M., Williams, K. D., Jones, A., & Zadro, L. (2008). The effects of
ostracism on self-regulation in the socially anxious. Journal of Social
and Clinical Psychology, 27, 471–504. doi:10.1521/jscp.2008.27.5.471
Ostafin, B. D., Marlatt, G. A., & Greenwald, A. G. (2008). Drinking
without thinking: An implicit measure of alcohol motivation predicts
failure to control alcohol use. Behavior Research and Therapy, 46,
1210–1219. doi:10.1016/j.brat.2008.08.003
Parasuraman, R. (1979, August 31). Memory load and event rate control
sensitivity decrements in sustained attention. Science, 205, 924–927.
Park, S. H., Glaser, J., & Knowles, E. D. (2008). Implicit motivation to
control prejudice moderates the effect of cognitive depletion on unin-
tended discrimination. Social Cognition, 26, 401–419. doi:10.1521/
Pocheptsova, A., Amir, O., Dhar, R., & Baumeister, R. F. (2009). Deciding
without resources: Resource depletion and choice in context. Journal of
Marketing Research, 46, 344–355.
Pottier, A., Bouckaert, J., Gilis, W., Roels, T., & Derave, W. (2010). Mouth
rinse but not ingestion of a carbohydrate solution improves 1-h cycle
time trial performance. Scandinavian Journal of Medicine & Science in
Sports, 20, 105–111. doi:10.1111/j.1600-0838.2008.00868.x
Reynolds, S., & Day, J. (1984, August). Monte Carlo studies of effect size
estimates and their approximation in meta-analysis. Paper presented at
the meeting of the American Psychological Association, Toronto, On-
tario, Canada.
Richeson, J. A., & Shelton, J. N. (2003). When prejudice does not pay:
Effects of interracial contact on executive function. Psychological Sci-
ence, 14, 287–290. doi:10.1111/1467-9280.03437
Richeson, J. A., & Trawalter, S. (2005). Why do interracial interactions
impair executive function? A resource depletion account. Journal of
Personality and Social Psychology, 88, 934–947. doi:10.1037/0022-
Richeson, J. A., Trawalter, S., & Shelton, J. N. (2005). African Americans’
implicit racial attitudes and the depletion of executive function after
interracial interactions. Social Cognition, 23, 336–352. doi:10.1521/
Rosenbaum, D. A., Carlson, R. A., & Gilmore, R. O. (2001). Acquisition
of intellectual and perceptual–motor skills. Annual Review of Psychol-
ogy, 52, 453–470. doi:10.1146/annurev.psych.52.1.453
Rosenberg, M. S. (2005). The file-drawer problem revisited: A general
weighted method for calculating fail-safe numbers in meta-analysis.
Evolution, 59, 464468. doi:10.1111/j.0014-3820.2005.tb01004.x
Sagie, A., & Koslowsky, M. (1993). Detecting moderators with meta-
analysis: An evaluation and comparison of techniques. Personnel Psy-
chology, 46, 629640. doi:10.1111/j.1744-6570.1993.tb00888.x
Sa´nchez-Meca, J., & Marı´n-Martı´nez, F. (1997). Homogeneity tests in
meta-analysis: A Monte Carlo comparison of statistical power and Type
I error. Quality and Quantity, 31, 385–399. doi:10.1023/A:
Sa´nchez-Meca, J., & Marı´n-Martı´nez, F. (1998). Testing continuous mod-
erators in meta-analysis: A comparison of procedures. Journal of Math-
ematical and Statistical Psychology, 51, 311–326.
Sansone, C., & Smith, J. L. (2000). The “how” of goal pursuit: Interest and
self-regulation. Psychological Inquiry, 11, 306–309. doi:10.1207/
Schmeichel, B. J. (2007). Attention control, memory updating, and emo-
tion regulation temporarily reduce the capacity for executive control.
Journal of Experimental Psychology: General, 136, 241–255. doi:
Schmeichel, B. J., & Baumeister, R. F. (2004). Self-regulatory strength. In
R. F. Baumeister & K. D. Vohs (Eds.), Handbook of self-regulation:
Research, theory and applications (pp. 84–98). New York, NY: Guild-
ford Press.
Schmeichel, B. J., Demaree, H. A., Robinson, J. L., & Pu, J. (2006). Ego
depletion by response exaggeration. Journal of Experimental Social
Psychology, 42, 95–102. doi:10.1016/j.jesp.2005.02.005
Schmeichel, B. J., & Vohs, K. D. (2009). Self-affirmation and self-control:
Affirming core values counteracts ego depletion. Journal of Personality
and Social Psychology, 96, 770–782. doi:10.1037/a0014635
Schmeichel, B. J., & Zell, A. (2007). Trait self-control predicts perfor-
mance on behavioral tests of self-control. Journal of Personality, 75,
743–755. doi:10.1111/j.1467-6494.2007.00455.x
Schmidt, K. H., Neubach, B., & Heuer, H. (2007). Self-control demands,
cognitive control deficits, and burnout. Work and Stress, 21, 142–154.
Schouwenburg, H. C. (2004). Academic procrastination: Theoretical no-
tions, measurement, and research. In H. C. Schouwenburg, C. H. Lay,
T. A. Pychyl, & J. R. Ferrari (Eds.), Counseling the procrastinator in
academic settings (pp. 3–17). Washington, DC: American Psychological
Schwarzer, R., & Jerusalem, M. (1995). Self-efficacy measurement: Gen-
eralized self-efficacy scale (GSES). In J. Weinman, S. Wright, & M.
Johnston (Eds.), Measures in health psychology: A user’s portfolio (pp.
35–37). Windsor, England: NFER-Nelson.
See, J. E., Howe, S. R., Warm, J. S., & Dember, W. N. (1995). Meta-
analysis of the sensitivity decrement in vigilance. Psychological Bulle-
tin, 117, 230–249. doi:10.1037/0033-2909.117.2.230
Seeley, E. A., & Gardner, W. I. (2003). The “selfless” and self-regulation:
The role of chronic other-orientation in averting self-regulatory deple-
tion. Self and Identity, 2, 103–117. doi:10.1080/15298860309034
Segerstrom, S. C., & Nes, L. S. (2007). Heart rate variability reflects
self-regulatory strength, effort, and fatigue. Psychological Science, 18,
275–281. doi:10.1111/j.1467-9280.2007.01888.x
Shamosh, N. A., & Gray, J. R. (2007). The relation between fluid intelli-
gence and self-regulatory depletion. Cognition & Emotion, 21, 1833–
1843. doi:10.1080/02699930701273658
Sheppes, G., Catran, E., & Meiran, N. (2009). Reappraisal (but not dis-
traction) is going to make you sweat: Physiological evidence for self-
control effort. International Journal of Psychophysiology, 71, 91–96.
Steel, P. (2007). The nature of procrastination: A meta-analytic and theo-
retical review of quintessential self-regulatory failure. Psychological
Bulletin, 133, 65–94. doi:10.1037/0033-2909.133.1.65
Stewart, C. C., Wright, R. A., Hui, S.-K. A., & Simmons, A. (2009).
Outcome expectancy as a moderator of mental fatigue influence on
cardiovascular response. Psychophysiology, 46, 1141–1149. doi:
Stillman, T. F., Tice, D. M., Fincham, F. D., & Lambert, N. M. (2009). The
psychological presence of family improves self-control. Journal of So-
cial and Clinical Psychology, 28, 498–529. doi:10.1521/
Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of
social behavior. Personality and Social Psychology Review, 8, 220–247.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions.
Journal of Experimental Psychology, 18, 643–661. doi:10.1037/
Stucke, T. S., & Baumeister, R. F. (2006). Ego depletion and aggressive
behavior: Is the inhibition of aggression a limited resource? European
Journal of Social Psychology, 36, 1–13. doi:10.1002/ejsp.285
Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High self-control
predicts good adjustment, less pathology, better grades, and interper-
sonal success. Journal of Personality, 72, 271–324. doi:10.1111/j.0022-
Tice, D. M., Baumeister, R. F., Shmueli, D., & Muraven, M. (2007).
Restoring the self: Positive affect helps improve self-regulation follow-
ing ego depletion. Journal of Experimental Social Psychology, 43,
379–384. doi:10.1016/j.jesp.2006.05.007
Tice, D. M., & Bratslavsky, E. (2000). Giving in to feel good: The place
of emotion regulation in the context of general self-control. Psycholog-
ical Inquiry, 11, 149–159. doi:10.1207/S15327965PLI1103_03
Tice, D. M., Bratslavsky, E., & Baumeister, R. F. (2001). Emotional
distress regulation takes precedence over impulse control: If you feel
bad, do it! Journal of Personality and Social Psychology, 80, 53–67.
Tops, M., Lorist, M. M., Wijers, A. A., & Meijman, T. F. (2004). To stress
or relax: Neurochemical aspects of activity and rest. Gedrag en Organi-
satie, 17, 32–42.
Trawalter, S., & Richeson, J. A. (2006). Regulatory focus and executive
function after interracial interactions. Journal of Experimental Social
Psychology, 42, 406412. doi:10.1016/j.jesp.2005.05.008
Tyler, J. M. (2008). In the eyes of others: Monitoring for relational value
cues. Human Communication Research, 34, 521–534. doi:10.1111/
Tyler, J. M., & Burns, K. C. (2008). After depletion: The replenishment of
the self’s regulatory resources. Self and Identity, 7, 305–321. doi:
Tyler, J. M., & Burns, K. C. (2009). Triggering conservation of the self’s
regulatory resources. Basic and Applied Social Psychology, 31, 255–
266. doi:10.1080/01973530903058490
Vohs, K. D. (2006). Self-regulatory resources power the reflective system:
Evidence from five domains. Journal of Consumer Psychology, 16,
217–223. doi:10.1207/s15327663jcp1603_3
Vohs, K. D., & Baumeister, R. F. (2004). Ego-depletion, self-control, and
choice. In J. Greenberg, S. L. Koole, & T. Pyszczynski (Eds.), Handbook
of experimental existential psychology (pp. 398410). New York, NY:
Guilford Press.
Vohs, K. D., Baumeister, R. F., & Ciarocco, N. J. (2005). Self-regulation
and self-presentation: Regulatory resource depletion impairs impression
management and effortful self-presentation depletes regulatory re-
sources. Journal of Personality and Social Psychology, 88, 632–657.
Vohs, K. D., Baumeister, R. F., Schmeichel, B. J., Twenge, J. M., Nelson,
N. M., & Tice, D. M. (2008). Making choices impairs subsequent
self-control: A limited-resource account of decision making, self-
regulation, and active initiative. Journal of Personality and Social Psy-
chology, 94, 883–898. doi:10.1037/0022-3514.94.5.883
Vohs, K. D., & Faber, R. J. (2007). Spent resources: Self-regulatory
resource availability affects impulse buying. Journal of Consumer Re-
search, 33, 537–547. doi:10.1086/510228
Vohs, K. D., & Heatherton, T. F. (2000). Self-regulatory failure: A
resource-depletion approach. Psychological Science, 11, 249–254. doi:
Wallace, H. M., & Baumeister, R. F. (2002). The effects of success versus
failure feedback on self-control. Self and Identity, 1, 35–42. doi:
Wallace, J. C., Edwards, B. D., Shull, A., & Finch, D. M. (2009). Exam-
ining the consequences in the tendency to suppress and reappraise
emotions on task-related job performance. Human Performance, 22,
23–43. doi:10.1080/08959280802540957
Wan, E. W., & Sternthal, B. (2008). Regulating the effects of depletion
through monitoring. Personality and Social Psychology Bulletin, 34,
32–46. doi:10.1177/0146167207306756
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-
dation of brief measures of positive and negative affect: The PANAS
scales. Journal of Personality and Social Psychology, 54, 1063–1070.
Webb, T. L., & Sheeran, P. (2003). Can implementation intentions help to
overcome ego-depletion? Journal of Experimental Social Psychology,
39, 279–286. doi:10.1016/s0022–1031(02)00527-9
Wegner, D. M., Schneider, D. J., Carter, S. R., & White, T. L. (1987).
Paradoxical effects of thought suppression. Journal of Personality and
Social Psychology, 53, 5–13. doi:10.1037/0022-3514.53.1.5
Westholm, A. (1987). Measurement error in causal analysis of panel data:
Attenuated versus inflated relationships. Quality and Quantity, 21, 3–20.
Wills, T. A., & Dishion, T. J. (2004). Temperament and adolescent sub-
stance use: A transactional analysis of emerging self-control. Journal of
Clinical Child and Adolescent Psychology, 33, 6981. doi:10.1207/
Wills, T. A., & Stoolmiller, M. (2002). The role of self-control in early
escalation of substance use: A time-varying analysis. Journal of Con-
sulting and Clinical Psychology, 70, 986–997. doi:10.1037/0022-
Wood, J. A. (2008). Methodology for dealing with duplicate study effects
in a meta-analysis. Organizational Research Methods, 11, 79–95. doi:
Wright, R. A. (2009). Sleep consistency as a mechanism for improving
inhibitory system strength. Stress and Health. Advance online publica-
tion. doi:10.1002/smi.1302
Wright, R. A., Junious, T. R., Neal, C., Avello, A., Graham, C., Herrmann,
L., Junious, S.,...Walton, N. (2007). Mental fatigue influence on
effort-related cardiovascular response: Difficulty effects and extension
across cognitive performance domains. Motivation and Emotion, 31,
219–231. doi:10.1007/s11031-007-9066-9
Wright, R. A., Martin, R. E., & Bland, J. L. (2003). Energy resource
depletion, task difficulty, and cardiovascular response to a mental arith-
metic challenge. Psychophysiology, 40, 98–105. doi:10.1111/1469-
Wright, R. A., Stewart, C. C., & Barnett, B. R. (2008). Mental fatigue
influence on effort-related cardiovascular response: Extension across the
regulatory (inhibitory)/non-regulatory performance dimension. Interna-
tional Journal of Psychophysiology, 69, 127–133. doi:10.1016/
Received April 30, 2009
Revision received February 16, 2010
Accepted February 19, 2010
... Here, people essentially sacrifice their initial preferences for large but delayed rewards (e.g., staying healthy into old age, waiting to earn two Marshmallows) in favor of immediately available but smaller rewards (e.g., skipping a training session to watch TV, eating one Marshmallow without waiting). Self-control is additionally affected by contextual factors like the prior exertion of mental effort in unrelated tasks (Giboin & Wolff, 2019;Hagger et al., 2010). For instance, people might find it more challenging to summon the self-control required for exercising in bad weather after an exhausting work day compared to a restful day of leisure (e.g., Schöndube et al., 2017). ...
... This temporary state of "ego depletion" ostensibly impairs the performance of subsequent self-control demanding behaviors by reducing an individual's ability to draw from the underlying resource. However, the empirical evidence for the assumptions of the strength model of self-control is inconsistent (e.g., Hagger et al., 2010Hagger et al., , 2016, the supporting literature is affected by publication bias (Wolff et al., 2018), and its mechanistic underpinnings have been questioned (Inzlicht et al., 2014). This encouraged the development of models that are more precise with respect to the mechanistic underpinnings of self-control. ...
... Advancing self-control research: Boredom as a potential confound The tight link between boredom and self-control has important implications for designing experiments that investigate situational fluctuations in self-control. These experiments commonly rely on a sequential two-task paradigm (e.g., Hagger et al., 2010), in which the initial task is designed to require either little self-control (e.g., transcribing a text without special rules; low-demand condition) or much self-control (e.g., transcribing a text while omitting certain letters; high-demand condition). The question then is whether these differences in self-control demands affect the performance in a subsequent self-control demanding task. ...
Full-text available
Self-control is a fragile mechanism by design, and there is an ongoing debate about the reasons for this fragility. A mechanistically plausible perspective is offered by reward-based models of self-control such as the expected value of control (EVC) theory. According to the EVC theory, the allocation of self-control requires effort, which serves as a dynamic signal that quantifies the costs of self-control. People thus use self-control sparingly and continuously adjust its allocation such that its expected value remains optimal. Boredom affects this process in two ways: It lowers the value of a focal activity compared to alternative activities and renders the maintenance of the focal activity more costly by increasing the required effort. Consequently, less control is allocated to an activity as it becomes boring, which in turn permits an optimal balance between goal-directed behavior (exploitation, directed exploration) and undirected exploration. The essential conclusion from the EVC perspective is that self-control must be fragile to be an adaptive function for behavior regulation. Conceiving self-control as a reward-based choice also helps to address some key open questions in the literature on boredom and self-control, such as the adequate conceptualization of trait boredom and the replicability of ego depletion effects.
... resources (for example, Hagger et al, 2010). These controversial findings suggest a finely nuanced understanding of the emotional journey in the self-starting process. ...
... Moreover, employees cannot focus enough on achieving task-related goals and lead to worse task performance when they are proactive actors at work. Even if they want to maintain their task performance, this will also require focal employees to take extra efforts and conduct more self-regulation, which would use up more resources and energy, and increase the ego-depletion (Hagger et al, 2010). According to the strength model of self-control (Hagger et al, 2010), depletion of individual regulatory resources is negatively related to mental health and well-being (Hülsheger and Schewe, 2011). ...
... Even if they want to maintain their task performance, this will also require focal employees to take extra efforts and conduct more self-regulation, which would use up more resources and energy, and increase the ego-depletion (Hagger et al, 2010). According to the strength model of self-control (Hagger et al, 2010), depletion of individual regulatory resources is negatively related to mental health and well-being (Hülsheger and Schewe, 2011). As such, engagement in proactive behaviour may potentially worsen employee mental health and personal well-being (Fay and Hüttges, 2017). ...
... Dans le cadre de nos activités professionnelles ou de loisir, il nous est tous un jour arrivé de ressentir de la fatigue cognitive pendant ou suite à la réalisation d'une tâche cognitive de longue durée qui demandait de l'effort mental. Cet état passager s'accompagne généralement d'une baisse de performance dans la tâche en cours ou dans celle qui suit et d'une sensation de manque d'énergie (Brown et al., 2020 ;Dang, 2018 ;Giboin & Wolff, 2019 ;Hagger, Wood, Stiff, & Chatzisarantis, 2010). ...
... Enfin, les méta-analyses qui ont pris en compte la durée de la tâche d'épuisement comme variable prédictive de la performance dans la tâche dépendante n'ont pas pu mettre en évidence d'effet significatif (Brown et al., 2020 ;Giboin & Wolff, 2019 ;Hagger et al., 2010). Ceci peut s'expliquer de deux manières. ...
Au quotidien nous réalisons des tâches cognitives qui requièrent de l’effort. Quand ces tâches sont particulièrement longues et/ou difficiles, elles induisent généralement de la fatigue cognitive. Cette dernière entraîne une baisse des performances qui s’accroît au cours du temps passé sur la tâche et s’observe dans des tâches subséquentes. Bien que ce champ de recherche soit très prolifique avec des centaines d’études publiées chaque année, l’existence de l’effet délétère de la fatigue cognitive sur la performance ainsi que les mécanismes explicatifs qui sous-tendent cet effet sont encore en débat. Cet article présente dans un premier temps la distinction entre le concept de fatigue cognitive et différents concepts proches tels que ceux de fatigue centrale, de fatigue physique ou encore d’ennui. Par la suite, les grandes théories explicatives de la fatigue cognitive sont abordées, ainsi que leurs divergences et convergences. Les méthodologies employées pour étudier la fatigue cognitive sont aussi explorées, ainsi que les variables confondantes qui doivent être contrôlées lors de la réalisation d’études sur ce sujet. Enfin, les perspectives pour les recherches futures sont explorées notamment les synergies entre les théories explicatives, les moyens de lutter contre la fatigue cognitive aiguë ou encore les pathologies qui peuvent accentuer la fatigue cognitive.
... In addition, self-control is a valuable resource. A metaanalysis of 83 studies found a significant effect on task performance (Hagger et al., 2010). It has also been shown that SRS are correlated with job performance and demographic factors including age, gender, and educational attainment. ...