Personality and Social Psychology Bulletin

http://psp.sagepub.com

Mechanisms of Self-Control Failure: Motivation and Limited Resources

Mark Muraven and Elisaveta Slessareva
Pers Soc Psychol Bull 2003; 29; 894
DOI: 10.1177/0146167203029007008
The online version of this article can be found at:
http://psp.sagepub.com/cgi/content/abstract/29/7/894

Published by:
http://www.sagepublications.com

On behalf of:

Society for Personality and Social Psychology, Inc.

Additional services and information for Personality and Social Psychology Bulletin can be found at:
Email Alerts: http://psp.sagepub.com/cgi/alerts
Subscriptions: http://psp.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

10.1177/0146167203253209
PERSONALITY
Muraven,
Slessareva
AND/SOCIAL
SELF-CONTROL
PSYCHOLOGY
FAILURE
BULLETIN

ARTICLE

Mechanisms of Self-Control Failure:

Motivation and Limited Resources
Mark Muraven
Elisaveta Slessareva
University at Albany
nally or internally determined) or delay gratification
(Barkley, 1997; Hayes, 1989; Shallice & Burgess, 1993).
Without self-control, the person would carry out his or
her normal, typical, or automatic behavior or engage in
immediate, short-term focused actions. For instance,
without self-control, a person would give in to temptation, quit when frustrated, disobey difficult instructions,
or otherwise follow his or her automatic, overlearned
patterns of behavior (Bargh & Chartrand, 1999). Someone who typically smokes and wants to stop must exert
self-control to break that habit. If the person does not
exert self-control, he or she will behave automatically
and smoke (Tiffany, 1990).
Self-control undoubtedly fails for a variety of reasons.
For example, a dieter may decide that that milkshake
ruined his diet so he loses nothing by bingeing (Herman
& Mack, 1975), or bad news may make the recovering
alcoholic decide that dealing with the negative mood is
more important than remaining abstinent (Tice,
Bratslavsky, & Baumeister, 2001). Individuals selfcontrol performance also may decline after exerting selfcontrol because self-control draws on a resource or
strength (self-control strength) that is limited and consumed in the process of exerting self-control (Muraven
& Baumeister, 2000). More specifically, if a resource or
strength is required for self-control, individuals who
have more of it should perform better on tasks that
require self-control than individuals lower in that
strength. And if that strength is consumed in the process

Research has found that individuals who are lower in selfcontrol strength because of previous self-control exertions perform more poorly on subsequent tests of self-control. The present
studies suggest that this effect may be moderated by motivation.
In particular, depletion and motivation jointly determine selfcontrol performance. Individuals who were depleted and
believed that the task would help others (Experiment 1) or
believed that their efforts could benefit them (Experiment 2) performed better on a subsequent test of self-control than individuals who were depleted and lower in motivation. The results of
Experiment 3 replicated these findings and suggested that depletion only affects performance on tasks that require self-control;
tasks that are difficult but do not require self-control are immune
to the effects of depletion. Hence, depleted individuals may compensate for their lack of self-control resources when sufficiently
motivated. The results may help explain the nature of self-control
strength.

Keywords: self-control; motivation; depletion; recourses

Self-control frequently fails despite the best intentions

of the person. For example, quitting smoking is not a

simple matter of wishing away an addiction. Smoking
cessation requires a great deal of effort to fight the urge
to smoke. Unfortunately, people often lose this battle
and start smoking again. To better understand and prevent these self-control breakdowns, researchers need to
know what causes these failures. In particular, are these
failures of self-control produced by an inability to exert
self-control, an unwillingness to put forth the effort to
succeed, or some combination of factors?
Self-control is the overriding or inhibiting of automatic, habitual, or innate behaviors, urges, emotions, or
desires that would otherwise interfere with goal-directed
behavior (Barkley, 1997; Baumeister, Heatherton, &
Tice, 1994; Kanfer & Karoly, 1972). People exert selfcontrol because they want to follow a rule (either exter-

Authors Note: This research is based in part on the first authors doctoral dissertation. He would like to thank the members of his committeeRoy Baumeister, Dianne Tice, Tim Curran, and Jagdip Singh
for their guidance. Correspondence concerning this article should be
addressed to Mark Muraven, Department of Psychology, University at
Albany, Albany, NY 12222; e-mail: [email protected].
PSPB, Vol. 29 No. 7, July 2003 894-906
DOI: 10.1177/0146167203253209
2003 by the Society for Personality and Social Psychology, Inc.

894
Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

Muraven, Slessareva / SELF-CONTROL FAILURE

of self-control, individuals level of self-control strength
will be temporarily diminished (their strength is
depleted) after an attempt at self-control. Thus, individuals who exert self-control will have less strength available for subsequent attempts at self-control. This suggests that individuals who exerted self-control should
perform more poorly on subsequent tests of self-control
as compared with individuals who did not exert selfcontrol.
As predicted by the self-control strength model,
research has found that self-control performance
declines after the exertion of self-control. For example,
individuals who controlled their thoughts were subsequently less able to suppress signs of amusement as compared with individuals who worked on multiplication
problems (Muraven, Tice, & Baumeister, 1998, Experiment 3). The math problem and thought suppression tasks
were equally frustrating, unpleasant, and difficult; the
two conditions only differed in how much self-control
they required. That is, one task was structured so that
there was a need to override a thought, whereas the
other condition was structured so that there was far less
conflict and inhibition needed to reach the experimental goal. Individuals who depleted their self-control
strength performed more poorly on subsequent tests of
self-control. These findings have been replicated several
times using a variety of self-control tasks (Baumeister,
Bratslavsky, Muraven, & Tice, 1998; for a review, see
Muraven & Baumeister, 2000; Muraven et al., 1998; Vohs
& Heatherton, 2000). The results of these studies suggest that subsequent self-control performance is directly
proportional to the amount of self-control the individual
previously exerted (Muraven, Collins, & Nienhaus,
2002).
Whether depletion invariably leads to poorer selfcontrol outcomes is less clear, however. For example,
individuals can compensate for a lack of sleep through
coffee and increased effort. Yet, most of us would prefer
our airline pilot had a good nights sleep rather than be
well caffeinated and highly motivated. The loss of some
resources cannot be compensated for. A cognitive load
or a shortfall in working memory is not easily overcome:
Enormous incentives cannot help an untrained individual learn strangers names while trying to remember a
10-digit sequence. Whether individuals can compensate
for a loss of self-control strength is even more unclear.
Despite extreme fatigue and stress after being shot down
during a reconnaissance mission over Bosnia, Captain
Scott OGrady was able to lie motionless while Serbian
search parties passed within feet of hima self-control
feat that people would find difficult under the best circumstances. On the other hand, during World War II,
radar operators fatigued and missed enemy targets
despite the deadly consequences of such inattention.

895

Thus, anecdotally, the evidence is mixed whether

individuals can overcome depletion. We are left with several important questions. Can a depleted addict in the
throes of withdrawal given his favorite drug resist the
temptation to use the drug if there is a gun to his head?
What if we replace the gun with the approval of a loved
one? Will a depleted student always give up studying
sooner than a nondepleted student, even if highly motivated? Prior research has found that individuals will
mobilize their energy when the incentives to do so are
sufficient but will fail to do so when the outcome is more
uncertain or less important (Brehm & Self, 1989; Wright
& Brehm, 1989). This research has not addressed selfcontrol strength, however. Understanding whether individuals can compensate for a loss of self-control strength
may provide insight into the nature of self-control
strength.
We undertook three experiments to explore whether
people can compensate for the loss of self-control
resources if given sufficient incentive (either internal or
external). If depleted individuals can compensate for
the loss of self-control resources, then giving them an
incentive to exert self-control should lead to better performance. If, on the other hand, resources alone determine self-control performance, then incentives to exert
self-control should have little effect on depleted participants (e.g., fear of death cannot motivate a sentry to pay
attention when he is mentally fatigued). Consistent with
the compensation hypothesis, we predicted that
depleted participants who were given little incentive to
exert self-control should perform more poorly than
depleted participants who were given greater incentives
and more poorly than nondepleted participants. The
amount of energy mobilized (motivation) should be a
product of incentives; in particular, the attractiveness of
the outcome and the perceived likelihood of success
(Brehm & Self, 1989; Wright & Brehm, 1989). Tasks that
are more attractive and attainable should increase effort;
less attractive tasks or tasks when success seems remote
should lead to diminished motivation. Depleted participants who were high in motivation may perform as well
as nondepleted participants on a test of self-control, thus
a focused contrast, rather than an interaction, should
have the greatest power for detecting the differences we
predict.
EXPERIMENT 1

Individuals may be more motivated when they believe

the outcome is important, even if the end result does not
benefit them directly (Batson, 1990). Thus, in Experiment 1, some of the participants were told that the
experimental procedures might help answer important
questions about memory that could eventually lead to
better treatments for Alzheimers disease. The other par-

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

896

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

ticipants were given the same instructions except that

the experimenter made no mention of Alzheimers disease or the potential benefits of the experiment. We reasoned that participants who thought that their efforts
could be potentially important and contribute to the
greater good would have a greater incentive to exert selfcontrol than individuals who were not told of potential
benefits to others.
If, indeed, individuals can compensate for the effects
of prior exertion of self-control (i.e., depletion), we predicted that participants who thought that the task was
unimportant and who were depleted should perform
more poorly on a subsequent test of self-control than
participants who thought that the task was important.
More broadly, self-control performance should depend
on the interaction of the individuals incentives to exert
self-control and their previous self-control exertions.
Individuals who exerted self-control in the first part of
the experiment should be in the same mood and no more
aroused than individuals who exerted less self-control in
the first part of the experiment. The depletion condition
should not differ from the nondepletion condition in
frustration, unpleasantness, or difficulty, either.
Method
Participants. A total of 43 (20 men and 23 women)
undergraduate students attending the University at
Albany participated in Experiment 1. Participants
received partial course credit for their participation.
Each participant was tested individually for a period of 1
hour.
Depletion phase. Participants were told that the purpose
of the study was to investigate the relationship between
cognitive ability and creativity and that during the course
of the experiment they would be asked to perform a cognitive task (a thought suppression task or a memory task)
and a creativity task that would involve working on two
puzzles.
Participants were told that they were going to work on
the cognitive task first. They were assigned to either the
thought suppression or the memory condition. Participants in the thought suppression condition were asked
to list any thoughts that came to their mind with the
admonition that they should avoid thinking about a
white bear. They were told that whenever they thought of
a white bear, they should write that thought down. The
experimenter emphasized that it was very important to
immediately change their thoughts and to try very hard
not to think of a white bear again. Participants in the
memory condition were instructed to memorize a list of
words provided to them by an experimenter. Although
memorizing a list of words may require overriding some
responses, memorizing should require far less inhibition

than suppressing a thought. Participants were given 5

min to work on the memory or thought suppression task.
Participants then were administered a manipulation
check to assess how difficult (e.g., How difficult was the
task?), pleasant (e.g., How unpleasant was the task),
and frustrating (e.g., How frustrating was the task?) the
previous task was. In addition, participants reported how
much self-control the instructions required (e.g., How
much were you fighting against an urge while working
on the task?). Participants responded to these questions
using a 30-point scale. To increase the reliability of the
manipulation checks, each theoretically important construct was assessed using multiple questions. In addition,
participants completed the Brief Mood Introspection
Scale (BMIS) (Mayer & Gaschke, 1988). The BMIS is a
well-validated and reliable instrument used to assess
mood valence and arousal. Participants rated their current feelings on 16 adjectives (e.g., happy, nervous) on a 7point scale. These 16 adjectives load on two mood factors: pleasantness-unpleasantness (valence) and arousalcalm (arousal). These mood factors have internal consistencies of .76 to .83 and well-established validity. The
BMIS helped to show that possible variations between
the groups on the second task could not be attributed to
differences in mood induced by the first task.
Self-control measurement phase. During the second part
of the experiment, participants were asked to work on
two puzzles that were introduced as a creativity measure.
This problem-solving task has been used in a number of
experiments to measure self-control, operationalized as
persistence in the face of frustration (Glass, Singer, &
Friedman, 1969; Muraven et al., 1998). The problemsolving task involved participants tracing two (unsolvable) geometric figures without retracing the same line
twice and without lifting a highlighter from the paper.
Once the participant solved a practice puzzle, the experimenter gave him or her the two unsolvable puzzles.
Before the tracing task was introduced, some participants were provided with an additional cover story that
emphasized the importance of the task. Those participants were told that the purpose of the study was to provide scientific evidence for the development of new therapies for patients with Alzheimers disease. Participants
in the other condition were told to try their best at the
task but were not told that their results may be used to
help persons with Alzheimers disease.
After participants signaled their desire to stop working on the puzzles (eight participants were stopped
because they exceeded the time limit of 45 min for the
experiment: three from suppress thoughts/important
condition, three from memory/important, and two
from memory/unimportant condition; none from the
thought suppress thoughts/unimportant condition
worked past 45 min), they were administered a manipu-

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

Muraven, Slessareva / SELF-CONTROL FAILURE

lation check. They were asked multiple questions about
their motivation to solve the puzzles (e.g., How motivated were you to solve the puzzles?). Finally, participants were thanked for their participation, debriefed
about the purpose of the experiment, and dismissed. All
participants were unaware that the puzzles were impossible and did not suspect that their performance on the
first task may have affected how long they persisted on
the second.
Results
Manipulation checks. Overall, the thought suppression
and memory conditions did not differ significantly on
frustration, difficulty, and unpleasantness, replicating
the previous studies. More specifically, for how difficult
participants found the first task (5 items, alpha = .76),
the thought suppression (M = 93.0) and memory (M =
98.5) conditions did not differ, t(41) = .63, ns. The conditions did not differ in pleasantness (6 items, alpha = .61),
t(41) = 1.08, ns (thought suppression, M = 99.9; memory,
M = 91.9) or frustration, either (3 items, alpha = .67),
t(41) = .79, ns (thought suppression, M = 58.0; memory,
M = 62.4).
On the BMIS, participants in the thought suppression
(M = 5.80) condition were in as pleasant a mood as participants in the memory (M = 7.89) condition, t(41) =
.57, ns. Likewise, participants in the thought suppression
(M = 26.64) condition were no more aroused than participants in the memory (M = 27.84) condition, t(41) =
.55, ns. Participants in the thought suppression condition (M = 17.24) were just as likely as participants in the
memory task (M = 14.53) to think that they did better
than most people (1 item), t(41) = 1.40, ns. A power analysis suggests that we had sufficient power to detect an
effect size of d = .4 (medium) approximately 70% of the
time, which indicates that we should have been able to
detect differences between conditions, had any existed.
In short, the results suggest that the initial tasks did not
differ in frustration, arousal, negative mood, or
difficulty.
The thought suppression (M = 111.47) and memory
task (M = 83.38) did differ in how much inhibition and
control over their selves participants reported exerting,
t(41) = 2.24, p < .05 (6 items, alpha = .92). That is, participants who were expected to be depleted did report
exerting more self-control than participants who were
not expected to be depleted.
The manipulation checks administered after the puzzles (dependent measure) indicated that participants
who thought that the task could potentially help people
with Alzheimers were more motivated than participants
who did not receive that instruction. In particular, participants in the important condition (M = 18.88) reported
being more motivated than participants in the unimpor-

897

tant condition (M = 9.50), t(41) = 3.12, p < .01 (3 items,

alpha = .68). This manipulation gave participants an
incentive to exert self-control.
Dependent measure. A focused contrast indicated that
participants in the unimportant condition who had to
suppress their thoughts quit working on the puzzles
much sooner than participants in the other three conditions, F(1, 39) = 4.36, p < .05 (see Figure 1). As predicted,
depleted participants given little incentive to exert selfcontrol performed the worst on a test of self-control.
When the eight participants who had to be stopped were
removed from the analysis, the focused contrast
remained significant, F(1, 31) = 4.49, p < .05.1 Likewise,
the analyses remained significant when the time spent
on the puzzles was transformed to a more normal
distribution.
Individual t tests showed that participants who suppressed their thoughts in the first part of the experiment
quit working much sooner when they believed the puzzles were unimportant (M = 20.5 s) than when they
believed the puzzles were important (M = 31.2 s), t(41) =
1.96, p < .05. When the task was important, the memory
(M = 27.1 s) and thought suppression (M = 31.2 s) conditions did not differ, t(41) = .68, ns. Of interest, depleted
individuals appear to be more sensitive to potential
rewards of the situation than nondepleted individuals,
an effect we found across all three experiments. When
the task was unimportant, participants who suppressed
their thoughts quit much sooner than participants who
did the memory task, t(41) = 1.91, p < .06, which basically
replicates the depletion effect demonstrated in previous
experiments (e.g., Muraven et al., 1998).
Discussion
Experiment 1 demonstrated that self-control performance is a product of individuals motivation to exert
self-control and their prior exertions of self-control. In
particular, people who are more depleted (i.e., exerted
self-control in the first part of the experiment) and who
were not given an incentive to exert self-control performed more poorly on a measure of self-control than
individuals who are depleted and who were given an
incentive and more poorly than individuals who are less
depleted. People can compensate for depletion if their
motivation is great enough.
The findings of Experiment 1 suggest that depletion
of self-control strength does not prevent the subsequent
exertion of self-control. Individuals can still exert selfcontrol when they are depleted, providing they are sufficiently motivated. In other words, in this experiment,
self-control broke down not because of a lack of
resources but instead because of a lack of resources and a
lack of motivation. Also, the lack of a main effect for
motivation and the similar performance of depleted and

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

898

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

35

Memory
Thought Suppression

30

Time On Puzzles
(minutes)

25

20

15

10

0
Not Important
Important
Outcome of Puzzles

Figure 1

Time spent working on impossible puzzles depends on first

task and importance of the puzzles.

nondepleted individuals in the high-motivation condition suggest that depleted individuals are more responsive to rewards than nondepleted individuals. There also
may an absolute ceiling in how much effort participants
are willing to put forth on (relatively) inconsequential
self-control tasks. Motivation may be a critical feature in
self-control performance. If that conclusion is supported in subsequent experiments, this finding has
important implications for why depletion leads to
poorer self-control performance.
Experiment 1 is noteworthy because we assessed
whether participants had to inhibit their behavior. The
results indicated that participants who had to suppress
their thoughts exerted more self-control than participants who had to memorize a list of words. That is not to
say that memorizing words (or any other task) does not
require inhibiting an impulse, only that in comparison
to suppressing a thought, memorizing words required
less self-control. Determining whether someone is exerting self-control is difficult (see Wegner & Pennebaker,
1993). Although experiments on self-control are typically designed to maximize the type of response conflicts
that require self-control (e.g., suppressing a thought),
assessing participants own experience helps to buttress
the argument that participants in the thought suppression condition were exerting self-control, whereas participants in the memory condition were not.
In addition, Experiment 1 is significant for its use of
multiple items in the manipulation check. This extensive manipulation check should have increased our abil-

ity to discern differences between conditions. Despite

this increase in power, the conditions did not differ in
mood, frustration, arousal, difficulty, or unpleasantness,
which is consistent with other studies on the aftereffects
of exerting self-control. Moreover, a power analysis suggested that we should have had enough power to find statistically significant differences had they existed. We
therefore suggest that the depletion and nondepletion
conditions did not differ in frustration, unpleasantness,
or other variables.
Similar to Experiment 1, Experiment 2 was designed
to investigate whether individuals can compensate for
depletion. That is, when motivated, do individuals who
previously exerted self-control perform at the same level
as individuals who did not previously exert self-control?
Experiment 2 used a different manipulation of motivation as well as different self-control tasks to examine the
generalizability of this effect and to examine potential
alternative explanations.
EXPERIMENT 2

People who perceive that success on a task is difficult

or unlikely may be less motivated (Brehm & Self, 1989;
Vroom, 1964; Wright & Brehm, 1989). Indeed, research
has found that motivation diminishes when the chance
of success seems remote (e.g., Wright, Brehm, &
Bushman, 1989). Hence, individuals may perform
extremely poorly on a test of self-control when they feel
that exerting self-control is unlikely to make a difference
in the eventual outcome. For example, a student may
believe that making sacrifices, such as resisting the temptation to party, may lead to a better grade. If the students
instructor has a reputation for being an unfair or arbitrary grader, however, the student may come to believe
that studying is irrelevant to the outcome. In that case,
the student may be likely to forego the library. The situation has been structured so that it undermines the students motivation to exert self-control. Expectations that
self-control is unlikely to succeed should reduce the
motivation to exert self-control. Moreover, we suggest
that if the student happens to be depleted, he may be
especially unlikely to go to the library, as compared to a
more rested student or a depleted student who is more
motivated. The outcome of a self-control task depends
on both motivation and self-control demands; individuals low in motivation and low in self-control strength are
the most likely to fail at self-control.
In Experiment 2, participants were asked to practice a
frustrating task. Overcoming frustration requires selfcontrol. Individuals whose self-control strength has been
depleted tend to quit frustrating tasks sooner than individuals whose strength was not depleted (Baumeister
et al., 1998; Muraven, 1998). Some of the participants

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

Muraven, Slessareva / SELF-CONTROL FAILURE

were told that practicing might make a difference in
their eventual performance on the game. Other participants believed that practicing would have little impact
on their final outcome. In other words, they believed
that the chance of success was remote, which should
reduce the amount of effort mobilized (Brehm & Self,
1989; Vroom, 1964; Wright & Brehm, 1989). Participants
who are lower in motivation should perform more
poorly on a test of self-control than participants higher
in motivation. This motivation should interact with the
previous self-control efforts to predict their self-control
performance on this frustrating task. In particular, participants who are depleted and working on what they
believe is a valueless task should quit the frustrating task
much sooner than depleted participants working on a
potentially useful and beneficial task and sooner than
nondepleted participants.
Method
Participants. Eighty-two undergraduate students (46
men and 36 women) attending Case Western Reserve
University were recruited for Experiment 2. They
received partial course credit in return for their participation. Each individual testing session lasted about 30
min.
Depletion phase. The experimenter explained to participants that they were taking part in an experiment concerning how mental organization affects performance.
For the first task, participants made a short speech on
how they would spend an ideal day that the experimenter recorded. Participants in the no instruction condition did not exert a great deal of self-control; they
spoke freely into the microphone. Participants in the
speech control condition had to exert more self-control
relative to the no instruction condition. More specifically, participants in the speech control were instructed
to avoid saying um or er during their speech. The use of
such speech fillers is relatively automatic and common
(see Christenfeld & Creager, 1996); therefore, suppressing these fillers should require a great deal of selfcontrol to override the habit of using them. Participants
in the speech control condition therefore should be
more depleted than participants in the no instruction
condition.
Self-control measurement phase. Participants then played
a frustrating game that involved rolling a ball around a
maze by tilting the playing surface while avoiding holes
in the path. Pretesting indicated that the game was very
frustrating and participants quickly tired of playing and
wished to quit. Previous research has shown that people
low in self-control tend to quit frustrating tasks sooner
(Baumeister et al., 1998; Glass et al., 1969; Muraven,

899

1998). Hence, how long participants played the game

was a measure of their self-control ability.
The experimenter told participants that they would
have a chance to practice the game before being tested
on it. Participants were told to ring a bell to indicate that
they were done practicing. As the experimenter was leaving the room for this practice session, he gave participants the motivational instructions. Participants in the
worthless practice condition were warned that the game
is very difficult and practice has little effect on final performance. Participants in the beneficial practice condition were told that although the game is very difficult,
with practice, their performance might improve. Thus,
participants in the worthless practice condition believed
that practicing was unlikely to help them, whereas participants in the beneficial practice condition anticipated a
task that could potentially help them succeed.
The experimenter then left the room and surreptitiously timed how long participants practiced the game.
When participants rang the bell to indicate that they
were done practicing (all participants stopped within 20
min; the experimenter did not terminate practice for
any participant), the experimenter returned to the
room and administered a series of manipulation checks.
Participants were then debriefed. No participant
reported being aware of the hypothesis, and participants
did not suspect that how long they were practicing the
task was being measured. Participants were similarly
unaware that the first task may have influenced how long
they worked on the second task.
Results
Manipulation checks. Using a 25-point scale, participants rated the speech control task as equally difficult
(M = 12.4) as the no instruction task (M = 12.7), t(80) =
.83, ns. Likewise, participants reported exerting the
same amount of effort on the speech control (M = 12.9)
and no instruction tasks (M = 12.0), t(80) = .631, ns. Participants also reported that the speech control task (M =
13.1) was as unpleasant as the no instruction task (M =
13.4), t(80) = .175, ns. Finally, participants liked the
speech control task (M = 12.4) and the no instruction
task (M = 11.0) equally, t(80) = 1.16, ns. The depletion
task did not differ from the nondepletion task except
the depletion task was designed to require the overriding of an urge or behavior, whereas the nondepletion
task did not.
Dependent measure. As we predicted, participants in the
speech control (depletion) condition who were told that
practicing the frustrating game would not improve their
performance performed more poorly than participants
in the other three conditions. As illustrated in Figure 2, a
focused contrast indicated that participants in the

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

900

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

400

experiment and who believed that practice was unlikely

to help quit practicing the frustrating task sooner than
participants who did not exert self-control initially and
sooner than participants who believed that practicing
could help their performance.

No Instructions
Speech Control

Seconds Spent Practicing

350
300

Discussion

250
200
150
100
50
0
Worthless

Beneficial

Beliefs about Outcome of Practicing

Figure 2

Time spent practicing a frustrating task depends on first

task and expectations of the outcome of practicing.

speech control (depletion) condition who were told that

practicing does not help improve performance quit
practicing sooner than did participants in the speech
control condition who were told that practicing does
lead to better performance and sooner than participants
in the no instruction condition, F(1, 78) = 5.39, p < .025.2
Individual t tests indicated that among participants in
the speech control condition, those who thought that
practice could lead to improved performance worked
on the frustrating task longer than those who thought
that practicing had little effect on final performance,
t(39) = 2.95, p < .01. Also, for the beneficial practice condition, the initial task (speech control, M = 341 s; no
instruction condition, M = 315 s) did not differ, t(39) =
.33, ns, but initial task did matter in the worthless practice condition (speech control, M =165 s; no instruction,
M = 242 s), t(39) = 1.98, p < .05. When low in motivation,
individuals previous self-control demands matter, but
when higher in motivation, individuals previous selfcontrol demands do not predict performance. We found
the same effect in Experiment 1. Self-control performance (time spent on the frustrating game) is a product
of motivation and previous self-control demands. We
also ran these analyses using an inverse square root transformation of the time spent practicing because we were
concerned that such practice data might be skewed. The
focused contrast remained significant, F(1, 78) = 5.45, p <
.01, as well as the individual t tests. In short, participants
who had to exert self-control in the first part of the

The results of Experiment 2 are consistent with the

results of Experiment 1. Depleted participants who
believed that practicing the self-control task was worthless performed more poorly than depleted participants
who believed the task could be beneficial. Motivation
and prior exertion of self-control jointly determined
performance so that depleted individuals low in motivation performed much more poorly than anyone else. In
summary, if sufficiently motivated, depleted individuals
may compensate for their loss of self-control strength.
Participants do not fail at self-control because they do
not have the necessary resources; they fail at self-control
because they compensate their loss of resources.
Both Experiments 1 and 2 suggest that motivation
and prior self-control demands interact to determine
subsequent self-control performance. Experiment 3 was
designed to extend these findings by examining
depleted individuals performance on tasks that require
self-control relative to their performance on tasks that
do not require self-control. That is, does depletion
reduce motivation on all tasks or are the effects of depletion specific to tasks that require self-control? The
answer to that question may provide insight into the
nature of self-control strength as well as help demonstrate the specificity of self-control resources.
EXPERIMENT 3

In Experiment 3, participants were paid based on

their self-control performance. Participants who are
paid more money for exerting self-control should be
more motivated to engage in self-control than participants who are paid less. In fact, incentives often do lead
to better self-control on various tests of self-control, such
as the cold-pressor task (Baker & Kirsch, 1991), coldclimate chamber tolerance (Johnson & Cabanac, 1983),
and isometric exercise persistence (Cabanac, 1986). In
short, money is a good incentive that can motivate individuals to exert more self-control. Paying participants
therefore affords a strong test of whether depleted individuals can overcome depletion when their motivation is
high or whether depletion leads to poorer self-control
performance regardless of the ultimate outcome. If, as
we expect, depletion contributes to but is not the ultimate cause of self-control failure, then depleted individuals who are paid to exert self-control should perform
just as well as nondepleted individuals.

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

Muraven, Slessareva / SELF-CONTROL FAILURE

There is a second question about the role of motivation in self-control that needs to be answered, however.
Perhaps depleted individuals are just lower in motivation
overall. That is, does depletion matter at all? Previous
studies have taken significant steps to consider alternative explanations; for instance, individuals who are low
in self-control strength will work longer on a task if quitting requires self-control (Baumeister et al., 1998, Experiment 4). Despite these findings, one could argue that
the exertion of self-control leads to negative moods or
frustration (Leith & Baumeister, 1996; Tice et al., 2001),
loss of self-efficacy (Bandura, 1997), learned helplessness (Seligman, 1975), or some other phenomenon that
results in poorer performance. One way to test these
alternative explanations for the results is to examine
depleted individuals performance on a task that does
not require self-control. The self-control strength model
predicts that the prior exertion of self-control should
affect performance on a task that requires self-control
but should have no effect on a task that does not require
self-control. Furthermore, if indeed individuals can
compensate for the effects of depletion, then motivation
and depletion should interact to predict performance
on a task that requires self-control, but the interaction
should not be significant for a task that does not require
self-control. The effects of depletion should be specific
to self-control tasks.
To test the specificity of self-control strength to selfcontrol tasks, participants consumed either a good-tasting
or bad-tasting beverage in Experiment 3. Drinking a
good-tasting beverage does not require overriding
strong, conflicting impulses and therefore should not
require much self-control. On the other hand, drinking
a bad-tasting beverage requires much more self-control
because the participant must override the natural desire
to stop performing an aversive action. Depletion should
reduce how much bad-tasting beverage participants consume, especially when they are low in motivation, but
should have no effect on how much good-tasting beverage they consume.
In summary, depleted participants who are poorly
paid for drinking a bad-tasting beverage should consume less than depleted participants who are well paid
for drinking a bad-tasting beverage and less than
nondepleted participants. Depleted participants who
are well paid for drinking a bad-tasting beverage should
consume as much as nondepleted participants. Depleted
participants should consume the same amount of goodtasting beverage as nondepleted participants, however.
More simply, the interaction between pay and depletion
should be significant for participants who consumed a
bad-tasting beverage but not for participants who consumed a good-tasting beverage.

901

Method
Participants. Ninety-seven (53 men and 44 women)
undergraduate students attending Case Western
Reserve University were recruited for Experiment 3.
They received partial course credit in return for their
participation. Each individual testing session lasted
about 30 min.
Depletion phase. The experimenter told participants
that they were taking part in an experiment looking at
how moods affect information processing. Participants
watched a 5-min video clip of a Robin Williams comedy
routine while their facial expressions were recorded.
Participants in the no instruction condition did not have
to override a behavior; they simply watched the video.
On the other hand, participants in the suppress reaction
condition were instructed to hide their emotional
expressions (i.e., laughing or smiling) while watching
the video. Regulating and inhibiting ones emotional
reactions should require self-control. Pretesting and
prior research (Muraven et al., 1998, Experiment 3)
have found that this clip is extremely funny and that suppressing ones desire to laugh does require self-control.
Therefore, participants in the suppress reaction condition should be lower in self-control strength than participants in the no instruction condition.
Self-control measurement phase. Following the depletion
task, participants were allowed to consume as much
orange-flavored Kool-Aid (a powdered instant drink
mix) as they wanted. Participants in the sweet beverage
condition drank Kool-Aid that was prepared normally,
with a cup of sugar added to the water. Participants in the
bitter beverage condition drank Kool-Aid that contained
a cup of vinegar rather than sugar. The bad-tasting beverage was rather bitter and unpleasant, although participants should have been able to tolerate it and even drink
it if they exerted enough self-control to override the normal desire not to consume bitter concoctions. The
experimenter recorded how much Kool-Aid participants consumed.
Participants in the high-pay condition were given 25
cents for every ounce of beverage they consumed. Participants in the low-pay condition were given 1 cent for
every ounce they consumed. Participants in the high-pay
condition were given a greater incentive to drink and
therefore should be more motivated to consume the beverage than participants in the low-pay condition.
After participants indicated that they had enough
Kool-Aid, the experimenter administered a manipulation check to examine differences in mood, motivation,
and liking of the task. Participants were then paid and
debriefed. Participants indicated no awareness of the
experimental hypothesis, and they did not suspect that

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

902

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

the first task might have influenced their performance

on the second task.
Results
Manipulation checks. Using a 25-point scale, participants in the high-pay condition reported being more
motivated (M = 18.50) to drink the beverage than participants in the low-pay condition (M = 13.33), t(95) = 4.97,
p < .0001. Also, participants in the high-pay condition
(M = 15.24) indicated that they worked harder at consuming the beverage than participants in the low-pay
condition (M = 11.91), t(95) = 2.29, p < .025. The pay
manipulation influenced how much effort participants
were willing to put forth.
Not surprisingly, participants rated the bitter beverage as more unpleasant (M = 5.17) than the sweet beverage (M = 17.14), t(95) = 12.47, p < .0001. However, participants reported that drinking the bitter beverage (M =
15.04) took as much effort as drinking the sweet beverage (M = 13.10), t(95) = 1.48, ns. This is consistent with
previous research that has found that self-control tasks
are not necessarily more effortful than tasks that do not
require self-control (e.g., Muraven et al., 1998).
Effortfulness is not a good measure of the amount of selfcontrol required. Instead, one must look to the amount of
overriding or inhibiting required. Drinking a bad-tasting
beverage should therefore require more self-control
than drinking a good-tasting one (consuming a goodtasting beverage will require self-control only if the person tries to override or exceed biological limits for liquid
capacity, which was highly unlikely in the present
experiment).
Dependent measure. The amount participants consumed was first analyzed in a 2 (taste) 2 (first task) 2
(pay) ANOVA. Understandably, there was a main effect
for taste, F(1, 89) = 43.13, p < .0001. Participants who
drank the bitter beverage consumed less than participants who drank the sweet beverage. The main effect for
first task, F(1, 89) = .12, ns, and pay, F(1, 89) = 2.13, ns, was
not significant, and neither were the two-way interactions between pay and taste, F(1, 89) = .001, ns; pay and
first task, F(1, 89) = .178, ns; and taste and first task, F(1,
89) = .134, ns. The three-way interaction between taste,
first task, and pay was significant, F(1, 89) = 5.51, p < .025.
To facilitate comprehension, we examined the simple
interaction effects (First Task Pay) for each level of
taste.
For the bitter beverage, there was a significant interaction between first task and pay, F(1, 42) = 4.34, p < .05, as
shown in Figure 3. More important, consistent with the
hypothesis that depleted individuals who are low in motivation will drink the least, a focused contrast suggested
that participants in the suppress reaction, low-pay condition consumed less than participants in any other condi-

tion (although the difference was marginally

significant), F(1, 42) = 3.22, p < .08.
Simple effects tests indicated that participants in the
suppress reaction (depletion) condition who were well
paid (M = 21.8) consumed more than participants in the
suppress reaction condition who were poorly paid (M =
9.09), t(20) = 2.19, p < .05. Participants could apparently
compensate for depletion if sufficiently motivated.
When poorly paid, participants who had to suppress
their reaction to the film consumed the same amount as
did participants in the no instruction condition (M =
17.5), t(20) = 1.44, ns. Participants in the suppress reaction condition who were well paid consumed as much as
participants in the high-pay, no instruction (no depletion) condition (M = 13.4), t(20) = 1.51, ns, and as much
as participants in the low-pay, no instruction condition,
t(20) = .75, ns. Although the high-pay, no instruction
condition appears discrepant, this cell does not differ
from the low-pay, no instruction condition, which is consistent with the previous experiments that found
nondepleted participants are less sensitive to motivational cues than depleted participants. In summary,
depleted participants who were given an incentive to
exert self-control performed as well as nondepleted
participants.
For the sweet beverage, a different pattern of results
emerged. As demonstrated by Figure 3, the interaction
between pay and first task was not significant, F(1, 47) =
1.71, ns. A focused contrast also indicated the participants in the suppress reaction condition who were
poorly paid (M = 34.9) consumed just as much as participants in the suppress reaction who were well paid (M =
33.6) and as much as participants in the no instruction
condition (high pay, M = 41.6; low pay, M = 31.3), F(1, 47)
= .013, ns. Depletion and motivation do not interact to
predict performance on a task that does not require selfcontrol. To summarize, participants consumed the same
amount of sweet beverage regardless of their level of
depletion. An individuals performance on a task that
does not require self-control, such as drinking a goodtasting beverage, was not affected by previous self-control
demands.
Alternative explanations. The differences between conditions in amount of bitter beverage consumed were
unrelated to mood, arousal, frustration, or any other
variable as measured at the end of the experiment. More
specifically, the amount of bitter beverage participants
consumed did not correlate with their perception of difficulty of the first task (the depletion task), r(45) = .13;
effort they exerted on the first task, r(45) = .052;
unpleasantness of the first task, r(45) = .015; and liking
of the first task, r(45) = .055, all ps > .35. These factors did
not mediate the relationship between initial self-control
task and final self-control performance. Furthermore,

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

Muraven, Slessareva / SELF-CONTROL FAILURE

sour

Kool-aid Drank (oz)

40

903

sweet

No Instructions
Suppress Reaction

30

20

10

high pay
low pay
Payment Condition

high pay
low pay
Payment Condition
Figure 3

Sweet and sour Kool-Aid consumed, based on first task and pay.

the two-way interaction between first task and pay for the
amount of bitter-tasting beverage consumed remained
significant when any of these variables were added as a
covariate to the ANOVA model to control for mood
effects, ps < .05. Differences in mood and effort exerted
in the first task did not influence how much participants
consumed. The lack of mood effects in the current
experiment is consistent with previous research on
depletion.
Discussion
Similar to the results of Experiments 1 and 2, the
results of Experiment 3 suggest that the incentives for
exerting self-control and level of self-control strength
(depletion) jointly determine the amount of self-control
exerted subsequently. If depleted participants were not
given a sufficient incentive to exert self-control, they performed more poorly than nondepleted participants.
Conversely, when motivated, depleted participants performed as well as nondepleted participants on a test of
self-control. The results suggest that depleted individual
are not unable, but perhaps unwilling, to exert self-control.
Motivation had an effect on depleted participants but no
effect on nondepleted participants, which suggest depletion may increase individuals sensitivity to the rewards
of the situation. Alternative explanations, such as mood
or difficulty of the first task, cannot easily account for the
results.
Also, consistent with prior research on depletion
(e.g., Muraven et al., 1998), Experiment 3 demonstrated
that the effects of prior exertion of self-control are lim-

ited to self-control tasks. Drinking a large quantity of a

good-tasting beverage was judged as equally difficult as
drinking a bitter beverage, yet depleted participants
performance on a task that required self-control was very
different from depleted participants performance on a
task that did not require self-control. The significant factor in depleted participants performance may be how
much self-control the task required, not how difficult the
task was. Self-control strength has no impact on tasks
that do not require self-control; the effects of depletion
apparently are specific to tasks that require self-control.
GENERAL DISCUSSION

Self-control performance may be a product of individuals previous exertions of self-control and their incentives for exerting self-control. Whereas previous studies
(Baumeister et al., 1998; Muraven et al., 1998) have
found that after inhibiting a strong impulse individuals
perform more poorly on subsequent tests of self-control,
the present studies demonstrated that that effect is moderated by motivation. In particular, depleted participants who are working on a self-control task that they
believe could help others (Experiment 1) persisted on a
frustrating task longer than depleted participants who
believed the task was unlikely to help others. Depleted
participants who worked on a task that they believed was
unlikely to benefit them (Experiment 2) also were less
likely to keep practicing as compared to depleted participants who believed that the task could benefit them, and
they were less likely to keeping practicing than
nondepleted individuals. Finally, Experiment 3 demon-

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

904

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

strated that when the incentives for exerting self-control

are small, depleted individuals are less likely to quaff an
unpleasant beverage than nondepleted participants, but
when the incentives are large, that difference disappears.
Combining the three experiments meta-analytically, we
found that contrast was significant, average d = .67, p <
.01. Across three experiments, depleted individuals
given little incentive to exert self-control performed
more poorly on tests of self-control than depleted individuals who were given greater incentives to exert selfcontrol and more poorly than nondepleted individuals.
Indeed, depleted individuals who were high in motivation performed as well as nondepleted individuals who
were high in motivation.
The results of Experiment 3 also suggest that exerting
self-control only affects subsequent performance on
tasks that require self-control and has no impact on tasks
that do not require self-control. The effects of depletion
seem to be specific to self-control. This finding suggests
that there may be something unique about exerting selfcontrol that leads to poorer self-control performance
subsequently. This conclusion is borne out by the manipulation checks. Across all three studies, a common conclusion was that the initial tasks did not differ significantly except in the amount of self-control they
demanded.3 More specifically, the initial tasks did not
differ in relative difficulty, frustration, effort required,
unpleasantness, or other variables. Participants who had
to exert self-control were in the same mood and were no
more aroused than participants who did not have to
exert self-control. Controlling for the difficulty, unpleasantness, or dislike of the initial task also had minimal
impact on the relationship among motivation, depletion, and self-control performance. Because mood, frustration, and unpleasantness were not correlated with
self-control performance and did not differ across conditions, it seems unlikely that the decline in self-control
perform is a product of these variables. Hence, theories
such as negative state relief, learned helplessness, or selfefficacy, which require these variables as mediators, are
hard pressed to explain the present findings. Moreover,
one would expect that learned helplessness would affect
all outcomes equally. Yet, in Experiment 3, we found that
outcomes that required self-control were affected by participants initial exertion of self-control, whereas outcomes that did not require self-control were not
affected. A factor specific to exerting self-control seems
to be related to poorer performance on subsequent tests
of self-control.
To summarize, the results imply that self-control performance is determined by more than just previous selfcontrol demands; motivation has a role. There are several potential explanations why motivation may compensate for the effects of depletion. First, the exertion of

self-control might lower individuals willingness to exert

self-control (effort or motivation). This explanation discounts the role of resources and just assumes that motivation is the critical factor. Although the motivation
explanation is potent, it fails to explain why the effects of
exerting self-control are specific to self-control tasks, as
demonstrated by Experiment 3. This account also is
hard pressed to explain why difficult tasks that do not
require self-control have no effect on subsequent selfcontrol performance. In short, there is something
unique to exerting self-control in that it only affects selfcontrol tasks. A general lack of motivation does not seem
like a viable explanation.
An alternative explanation is that the exertion of selfcontrol reduces motivation specific to self-control.
There may be two reasons for this, both of which spring
from the idea that self-control acts like a limited
resource. First, after exerting self-control, subsequent
attempts at self-control may be more difficult and
require more effort. Much like a fatigued person will
find digging a ditch more demanding than a rested person, a depleted person may have to work harder to exert
self-control. Motivation increases the likelihood of putting forth the effort. We call this the well model because
much like water becomes harder to obtain as a well dries
up, self-control may become more difficult as strength is
depleted. This well model supposes that self-control is
indeed a limited resource.
The second potential model that explains the current
findings assumes that depleted individuals are motivated
to conserve strength. Similar to the well model, this
model assumes that exerting self-control depletes a limited resource, but instead, this model assumes that selfcontrol become less likely (not more difficult) as
strength decreases. In particular, individuals should be
motivated to conserve limited resources. Moreover, this
motivation to conserve should be greater in individuals
low in resources (such as depleted individuals) than in
individuals flush with resources. Much like $100 is worth
more to a starving student than to a millionaire, selfcontrol strength may be more valuable to a depleted person than a well-rested person. Hence, self-control breaks
down after the exertion of self-control not because individuals become unable to exert self-control or because
self-control becomes more difficult but because they
wish to conserve strength. This is not necessarily a conscious, deliberative process but rather something individuals do continually with very little awareness. Depletion may cause people to become more selective in
whether they will exert self-control. Indeed, in the present experiments, manipulations of motivation had a
greater effect on the self-control performance of
depleted individuals than nondepleted individuals (but,
consistent with this model, Experiment 3 demonstrated

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

Muraven, Slessareva / SELF-CONTROL FAILURE

that manipulations of motivation had the same effect on
depleted and nondepleted individuals when the task did
not require self-control). Increasing individuals motivation to exert self-control gives them incentive to use their
strength and overcome their natural desire to conserve
strength. We call this the conservation model.
Future research should be able to differentiate
between the conservation and the well model by manipulating participants expectations for future self-control
demandsthe conservation model suggests that individuals should be sensitive to future self-control
demands, whereas the well model suggests that individuals will not be. Both models are mute, however, on
whether individuals can compensate for more extreme
levels of depletion. The present set of studies suggest
that people can compensate for a small loss of strength.
Individuals may not be able to compensate for a greater
depletion of strength, much like there are physiological
limits on human performance, however. If very
depleted, a dieter may be unable to resist the urge to eat,
no matter how strong his motivation may be. Motivation
can compensate for but not eliminate depletion. The
present findings also do not speak to whether there may
be irresistible urgesurges so overwhelming that no
amount of motivation and strength are enough to resist
them.
Although the self-control strength model can account
for the results, it is unclear whether such a limited
resource model is necessary to explain the results. However, the results of these and other studies (e.g.,
Baumeister et al., 1998; Muraven et al., 1998, 2002)
clearly demonstrate that the exertion of self-control
leads to a subsequent decline in self-control performance. Furthermore, as noted above, the decrease in
self-control performance among depleted individuals is
not produced by negative moods, arousal, frustration, or
physical fatigue and is specific to self-control tasks. Thus,
we suggest that people act as if self-control is a limited
resource. Whether a resource truly underlies self-control
and, if so, the exact nature of that resource remains to be
investigated. At this point, however, we can say with reasonable confidence that the self-control strength model
seems to capture the essence of self-control.
Unlike prior studies (e.g., Baumeister et al., 1998;
Muraven et al., 1998) that have held motivation constant, both motivation and depletion were manipulated
in the current studies. When motivation was low, we replicated previous results that found that the exertion of
self-control leads to poorer self-control performance.
When motivation is high, the effects of depletion disappear, however. We did not find a main effect for motivation, perhaps because of a ceiling in how much effort
participants are willing to put forth in a laboratory study
of self-control (which is also consistent with the conser-

905

vation model). These findings suggest that individuals

can compensate for the loss of self-control strength. Individuals should be most likely to suffer a breakdown in
self-control when they are depleted and low in motivation. Similarly, a depleted individual is likely to stop controlling unimportant behaviors before loosening
restraints over more important behaviors. This may
explain why participants in studies on depletion quit
frustrating tasks without cursing the experimenter aloud
for the pain they have suffered.
The finding that individuals can compensate for a loss
of self-control resources with increased motivation may
be important for other models that posit limited
resources. For example, Gilbert, Tafarodi, and Malone
(1993) suggest that when individuals energy is being
reduced through a cognitive load they are less able to
correct for false ideas. The present study suggests that
this may only be true when individuals are low in motivation; when their motivation is high, individuals who are
cognitively busy may be able to debias themselves as well
as individuals who are not as overloaded. Similarly, one
would expect that individuals low in cognitive resources
but high in motivation would be more able to make mental corrections (Wilson & Brekke, 1994) than individuals
low in resources and low in motivation.
The results also suggest that an understanding of
motivation is crucial to understanding self-control.
Thus, various models of motivation (Brehm & Self, 1989;
Locke & Latham, 1990) may help to explain self-control
failures, particularly when individuals are low in selfcontrol strength. Most notably, we did not measure or
differentiate between intrinsic and extrinsic motivation
(Deci & Ryan, 1985). Superficially, the motivators in the
present study ranged from highly internal (Experiment
1) to highly external (Experiment 3), and all had the
same effect on depletion. This suggests that, at least for
depleted individuals, the type of incentive for exerting
self-control is not crucial. Future researchers may want
to look at this distinction in more detail to determine if
motivation orientation plays a role in depletion.
Emotions, in particular positive affect, may be important in compensating for the effects of depletion. Prior
research has found that positive affect may lead to
increased motivation because positive moods may signal
that ones resources are adequate and therefore a small
loss of resources or costs is acceptable (Aspinwall, 1998;
Trope & Pomerantz, 1998). Negative emotions may have
the opposite effect on the willingness to use and lose
resources. Thus, the extent to which a manipulation of
motivation affects individuals mood may alter their willingness to lose self-control resources. Hence, mood may
affect self-control outcomes.
In conclusion, self-control performance is a product
of the individuals motivation to exert self-control and

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008

906

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

his or her resources. Individuals can compensate for a

lack of self-control resources with increased motivation,
at least up to a point. Whereas a lack of resources may
contribute to self-control failure, it is not the ultimate
cause. The moderating influence of motivation suggests
that depletion does not necessarily lead to self-control
failure. The role of motivation in self-control also can
help explain why, when depleted, people may lose control of their appetites but not their temper (or control
their appetites and lose emotional control). In short,
motivation and resources are both important components of self-control.
NOTES
1. The interaction between initial task and incentive condition
approached significance for the full sample, F(1, 39) = 3.28, p < .07, and
was significant for the subsample of participants who were not stopped,
F(1, 31) = 5.19, p < .05.
2. The unfocused interaction between initial task and practice condition was not significant, F(1, 89) = .089, ns.
3. That is not to say that one task required self-control and the other
did not. Instead, the tasks were structured so that one task should
require much more self-control than the other. Individuals reports of
the amount of inhibition and overriding required in Experiment 1 support that assumption.
REFERENCES
Aspinwall, L. G. (1998). Rethinking the role of positive affect in selfregulation. Motivation and Emotion, 22, 1-32.
Baker, S., & Kirsch, I. (1991). Cognitive mediators of pain perception
and tolerance. Journal of Personality and Social Psychology, 61, 504510.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York:
Freeman.
Bargh, J. A., & Chartrand, T. L. (1999). The unbearable automaticity
of being. American Psychologist, 54, 462-479.
Barkley, R. A. (1997). ADHD and the nature of self-control. New York:
Guilford.
Batson, D. (1990). How social an animal? The human capacity for caring. American Psychologist, 45, 336-346.
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.
Baumeister, R. F., Heatherton, T. F., & Tice, D. M. (1994). Losing control: How and why people fail at self-regulation. San Diego, CA: Academic Press.
Brehm, J. W., & Self, E. (1989). The intensity of motivation. Annual
Review of Psychology, 40, 109-131.
Cabanac, M. (1986). Money vs. pain: Experimental study of a conflict
in humans. Journal of the Experimental Analysis of Behavior, 46, 37-44.
Christenfeld, N., & Creager, B. (1996). Anxiety, alcohol, aphasia, and
ums. Journal of Personality and Social Psychology, 70, 451-460.
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination
in human behavior. New York: Plenum.
Gilbert, D. T., Tafarodi, R. W., & Malone, P. S. (1993). You cant not
believe everything you read. Journal of Personality and Social Psychology, 65, 221-233.
Glass, D. C., Singer, J. E., & Friedman, L. N. (1969). Psychic cost of
adaptation to an environmental stressor. Journal of Personality and
Social Psychology, 12, 200-210.

Hayes, S. C. (1989). Rule-governed behavior: Cognition, contingencies, and

instructional control. New York: Plenum.
Herman, C. P., & Mack, D. (1975). Restrained and unrestrained eating. Journal of Personality, 43, 647-660.
Johnson, K., & Cabanac, M. (1983). Human thermoregulatory behavior during a conflict between cold discomfort and money. Physiology and Behavior, 30, 145-150.
Kanfer, F. H., & Karoly, P. (1972). Self-control: A behavioristic excursion into the lions den. Behavior Therapy, 3, 398-416.
Leith, K. P., & Baumeister, R. F. (1996). Why do bad moods increase
self-defeating behavior? Emotion, risk taking, and self-regulation.
Journal of Personality and Social Psychology, 71, 1250-1267.
Locke, E. A., & Latham, G. P. (1990). A theory of goal setting and task performance. Englewood Cliffs, NJ: Prentice Hall.
Mayer, J. D., & Gaschke, Y. N. (1988). The experience and metaexperience of mood. Journal of Personality and Social Psychology, 55,
102-111.
Muraven, M. (1998). Mechanisms of self-control failure: Motivation and
limited resources. Unpublished doctoral dissertation, Case Western
Reserve University, Cleveland, OH.
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of limited resources: Does self-control resemble a muscle?
Psychological Bulletin, 126, 247-259.
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., 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.
Seligman, M. P. (1975). Helplessness: On depression, development, and
death. San Francisco: Freeman.
Shallice, T., & Burgess, P. (1993). Supervisory control of action and
thought selection. In A. Baddeley & L. Weiskrantz (Eds.), Attention: Selection, awareness, and control (pp. 171-187). Oxford, UK:
Oxford University Press.
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.
Tiffany, S. T. (1990). A cognitive model of drug urges and drug-use
behavior: Role of automatic and nonautomatic processes. Psychological Review, 97, 147-168.
Trope, Y., & Pomerantz, E. M. (1998). Resolving conflicts among selfevaluative motives: Positive experiences as a resource for overcoming defensiveness. Motivation and Emotion, 22, 53-72.
Vohs, K. D., & Heatherton, T. F. (2000). Self-regulatory failure: A
resource-depletion approach. Psychological Science, 11, 243-254.
Vroom, V. H. (1964). Work and motivation. New York: John Wiley.
Wegner, D. M., & Pennebaker, J. W. (1993). Changing our minds: An
introduction to mental control. In D. M. Wegner & J. W.
Pennebaker (Eds.), Handbook of mental control (pp. 1-12).
Englewood Cliffs, NJ: Prentice Hall.
Wilson, T. D., & Brekke, N. (1994). Mental contamination and mental
correction: Unwanted influences on judgments and evaluations.
Psychological Bulletin, 116, 117-142.
Wright, R. A., & Brehm, J. W. (1989). Energization and goal attractiveness. In L. A. Pervin (Ed.), Goal concepts in personality and social psychology (pp. 169-210). Hillsdale, NJ: Lawrence Erlbaum.
Wright, R. A., Brehm, J. W., & Bushman, B. J. (1989). Cardiovascular
responses to threat: Effects of the difficulty and availability of a
cognitive avoidance task. Basic and Applied Social Psychology, 10,
151-171.
Received February 12, 2002
Revision accepted August 7, 2002

Downloaded from http://psp.sagepub.com by Constantin Ticu on October 13, 2008