Competitively Contingent Rewards and Intrinsic Motivation: Can Losers Remain Motivated?

Abstract

We explored the effects on intrinsic motivation and ego-involved persistence of winning versus losing a competitively contingent reward and, for losers, the additional effects of receiving either positive performance feedback or performance-contingent rewards. Winners were more intrinsically motivated than losers. Losers given an explicit normative standard who received positive feedback for meeting the standard were more intrinsically motivated than losers who did not receive the additional standard and feedback. Losers who received a performance-contingent reward for reaching the same explicit standard displayed less intrinsic motivation behaviorally assessed than did losers who got positive feedback, but the two groups did not differ on self-reported enjoyment. Effects on enjoyment were mediated by perceived competence, but effects on free-choice behavior were not. People who lost the competition showed more ego-involved persistence than people who won or did not compete.

Full text

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Motivation and Emotion, Vol. 27, No. 4, December 2003 ( C
°2003)
Competitively Contingent Rewards and Intrinsic
Motivation: Can Losers Remain Motivated?1
Maarten Vansteenkiste2,4and Edward L. Deci3,4
We explored the effects on intrinsic motivation and ego-involved persistence of
winning versus losing a competitively contingent reward and, for losers, the addi-
tional effects of receiving either positive performance feedback or performance-
contingent rewards. Winners were more intrinsically motivated than losers. Losers
given an explicit normative standard who received positive feedback for meeting
the standard were more intrinsically motivated than losers who did not receive the
additional standard and feedback. Losers who received a performance-contingent
reward for reaching the same explicit standard displayed less intrinsic motivation
behaviorally assessed than did losers who got positive feedback, but the two groups
did not differ on self-reported enjoyment. Effects on enjoyment were mediated by
perceived competence, but effects on free-choice behavior were not. People who
lost the competition showed more ego-involved persistence than people who won
or did not compete.
KEY WORDS: competition; rewards; intrinsic motivation.
Competition between individuals or teams is a central aspect of most sports and of
many other life activities in our modern culture. In some competitive situations,
people’s goal is simply to win either because winning is enjoyable and exiciting for
them or, alternatively, because it helps them bolster their fragile egos. In other com-
petitive situations, people’s goal may include not only winning but also obtaining
1This research was supported by grants from the National Institute of Mental Health (MH-53385) and
the Fund for Scientific Research, Flanders, Belgium.
2Department of Psychology, University of Leuven, Tiensestraat, Belgium.
3Department of Psychology, University of Rochester, Rochester, New York.
4Address all correspondence either to Maarten Vansteenkiste, Department of Psychology, University
of Leuven, Tiensestraat 102, 3000, Belgium; e-mail: maarten.vansteenkiste@psy.kuleuven.ac.be; or
to Edward L. Deci, Department of Psychology, University of Rochester, Rochester, New York 14627;
e-mail: deci@psych.rochester.edu.
273
0146-7239/03/1200-0273/0 C
°2003 Plenum Publishing Corporation

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
274 Vansteenkiste and Deci
additional outcomes such as symbolic or monetary rewards that have been made
contingent upon winning. It is common, for example, to see professional athletes
compete in tournaments where the purse runs into tens of thousands of dollars,
and the real plum for winning may be product endorsements that yield many times
the value of the purse.
Remarkably, financial inducements have even made their way into amateur
athletics. Some parents offer financial incentives to their children for winning
at sports or other competitive activities, and the use of incentives for amateurs
has, to some extent, become institutionalized. Many college athletes receive large
scholarships based only on their athletic accomplishments and potentials, and
further, for example, in some amateur Belgian soccer leagues, players as young
as 16–17 years of age can earn monetary rewards equivalent to about $100 when
their team wins a game.
Financial incentives are not the only source of pressure on individuals to win
competitions.Parentsofevenveryyoungathletesmaybecomehighlyego involved
in the outcome of their children’s little league or high-school games, pressuring
the children to try harder and, perhaps, to win at whatever cost. As well, fans and
televisions audiences can add to the stress that some college athletes feel to win
big games. It seems that, for some people, playing well is not enough.
Research on Competition, Rewards, and Intrinsic Motivation
Given the importance in modern culture of competition and rewards made
contingent upon winning a competition, studies have begun to examine the ef-
fects of these factors on people’s intrinsic motivation for the activity at which they
compete. The following gives an overview of the types of research that have been
done.
When people compete against each other, either individually or as teams,
doing the same activity at the same time with each side trying to win, it is referred
toasdirectcompetition.Thecurrent study involveddirectcompetition and previous
competition studies that are herein reviewed were also of that type. In experiments
of direct competition, the three most basic between-group comparisons that can
be made are (1) winners can be compared to losers; (2) either winners or losers
can be compared to participants in a control group where participants did the same
activity alone, without competing and without any performance feedback (i.e., a
no-competition/no-feedback control group); and (3) either winners or losers can
be compared to participants in a control group where participants worked in the
presence of another person (typically an experimental accomplice) who did the
sameactivity with participantsbeing told simply todotheir best, with nomentionof
trying to win (i.e., a no-competition/implicit-feedback control group). In this third
case, the control-group participants get feedback in the sense that they see (or are
told) which participant did the activity more quickly (or, in some other way, did it
better).Using the former type of control groupallowsinvestigatorsto determine the

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 275
combined effects of the competition and the competitive outcome (e.g., competing
and winning), whereas the latter type of control group allows investigators to
determine the effects of the competition independent of the competitive outcome.
Studies of competition can be further complicated by the addition of rewards.
For example, as is frequently done in the real world, participants can be told that
the winner of a competition will receive a reward. Such rewards are referred to
as competitively contingent rewards (Ryan, Mims, & Koestner, 1983), and in real
situations with competitively contingent rewards some people are winners and
some are losers. Thus, with competitively contingent rewards, the participants
(a) compete, (b) have a competitive outcome—that is, they either win or lose the
competition, and (c) either receive or do not receive the desired reward.
Competitively contingent rewards differ from performance-contingent
rewards (Ryan et al., 1983), which is another relevant reward contingency, in the
following way. Performance-contingent rewards are given to people for doing well
atatask. This contingency is frequently instantiated by telling participants thatthey
will get a reward if they perform well according to some normative standard—for
example, if they perform better than 80% of the other participants who have done
the task. Thus, whereas competitively contingent rewards are given for beating an
opponent in a direct competition, performance-contingent rewards are given for
doing better than an implicit or explicit normative standard. Both a competitively
contingent reward condition and a performance-contingent reward condition can
be compared to either (a) controls group that do not get feedback comparable to
that which is implicit in winning (or losing) a competitively contingent reward
or in receiving (or not receiving) a performance-contingent reward; or (b) control
groups that do get either expected or unexpected feedback that is comparable to
the feedback that reward-group participants received implicitly.
An additional complexity of the competition and reward studies is that the
interpersonal context within which people compete or get rewards can either be
relatively pressuring and controlling or relatively nonpressuring and informational
(Deci & Ryan, 2000). That is, the individuals administering the competition or
rewards can pressure people either to win or to do well enough to get a reward, or
the individuals can be relatively nonpressuring, giving people the opportunity to
choose for themselves how hard to try to win or to earn a reward.
Finally, studies of intrinsic motivation have employed two different measures
of intrinsic motivation as a dependent variable. Some have used the so-called
free-choice measure in which, following the experimental period, participants are
left alone in the experimental room with the freedom to either do more of the
target activity or to do alternative activities, including just day dreaming. The
amount of time they spend with the activity is generally considered a measure of
their intrinsic motivation (Deci, 1971). The second assessment approach uses self-
reports. Because intrinsic motivation for a target activity is assumed to be based
in the interest people have for that activity, people’s reports of how interesting
and enjoyable they find the activity is also used as a measure of their intrinsic

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
276 Vansteenkiste and Deci
motivation. In the literature review that follows, in all cases when we refer to the
effect of some independent variable on intrinsic motivation, the effect has been
found with the free-choice behavioral measure (and in some cases, also with the
self-report measure) unless we say otherwise.
There have been a number of studies relating competition (either with or
without rewards) to intrinsic motivation, and four findings have emerged that are
directly relevant to the formulation of the current study.
First, a study by Deci, Betley, Kahle, Abrams, and Porac (1981) found that
when two individuals worked in the presence of each other on the same puzzle-
solving activity, those participants who had been told to try to beat their opponent
at solving the puzzles displayed less subsequent intrinsic motivation for the puzzle-
solving activity than did those who had simply been told to do their best and solve
the puzzles as quickly as they could, with no mention being made of winning or do-
ing better than the other person. Importantly, in that study, participants in the com-
petitive group won the competition and participants in the comparison group got
the same implicit positive feedback because they could see that they had solved the
puzzles faster than the other participant, who was an experimental accomplice. As
such,the comparison group was a no-competition/positive-feedbackcontrol group,
so the conclusion was that winning a direct, face-to-face competition decreased
intrinsic motivation relative to doing the same task in the presence of the other
and receiving implicit positive feedback about one’s performance. In short, when
working in the presence of another and receiving positive performance information
were held constant, competition undermined intrinsic motivation, suggesting that
competition itself, independent of the competitive outcome, is controlling.
Thereissome recent evidence that personality factorssuchaslevel of achieve-
ment motivation may, to some extent, moderate the detrimental effect on intrin-
sic motivation of direct competition (Epstein & Harackiewicz, 1992; Tauer &
Harackiewicz, 1999). However, the detrimental main effects has been replicated
by various researchers (e.g., Vallerand, Gauvin, & Halliwell, 1986) and the mod-
eration effect has been found only with self-reports of enjoyment. Thus, it does
appear that there is a detrimental main effect for competition when the information
implicit in the competitive outcome is held constant.
The second relevant finding was that the interpersonal climate within which a
competition occurred moderated the competition’s effects on intrinsic motivation
(Reeve & Deci, 1996). Specifically, when participants won a competition within
a controlling, pressuring interpersonal context, their intrinsic motivation was un-
dermined relative to that of participants who won the competition in a context
that did not pressure them to beat the opponent. Thus, a pressuring interpersonal
context has a negative main effect on intrinsic motivation when competition and
the competitive outcome are held constant. Further, the Reeve and Deci study in-
troduced a no-competition/no-feedback control group where participants worked
alone rather than in the presence of another. Results showed that winning a com-
petition in a context that was not pressuring enhanced intrinsic motivation relative

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 277
to a no-competition/no feedback control group, whereas winning the competition
in a pressuring context resulted in the same level of intrinsic motivation as being
in the no-competition/no-feedback control group.
When the Reeve and Deci (1996) study is considered together with the Deci
et al. (1981) study, the findings indicate that winning a competition tends to under-
mine intrinsic motivation relative to a no-competition/positive-feedback control
group where the presence of another and the positive information implicit in win-
ning a competition are held constant. Further, competing in a pressuring interper-
sonal context can make the experience even more controlling. However, winning a
competition does not undermine intrinsic motivation relative to not competing and
notreceivingany positivefeedback; in fact, winning a competition can lead to more
intrinsic motivation than this no-competition/no feedback control condition if the
competition occurs in a nonpressuring context. In short, then, competition itself
(independent of the information contained in the competitive outcome) tends to
undermine intrinsic motivation; positive competence feedback contained in win-
ning or doing well tends to enhance intrinsic motivation; and the interpersonal
context can influence which effect is more dominant.
The third relevant finding is that participants who lost a competition had less
subsequent intrinsic motivation than those who won (Reeve, Olson, & Cole, 1985;
Vallerand& Reid, 1984). In other words, when competition itself was held constant
across conditions, the competitive outcome (i.e., the information contained in
winning vs. losing) significantly affected intrinsic motivation. Further, compared
to a no-competition/no-feedback control group, those who lost a competition in
a nonpressuring setting showed diminished intrinsic motivation whereas those
who won the competition in a nonpressuring setting showed enhanced intrinsic
motivation (Reeve & Deci, 1996). Thus, positive competence feedback inherent
in winning tends to enhance intrinsic motivation whereas negative competence
feedback inherent in losing tends to diminish it relative to a no-competition/no-
feedback control condition.
The fourth relevant finding was that, when participants within a group setting
competed against the other members of the group to win a tangible monetary
reward, the average intrinsic motivation of all participants was less than that of the
control group members who did not compete for a reward (Pritchard, Campbell,
& Campbell, 1977). In other words, when the winners and losers were considered
together, these competitively-contingent rewards undermined intrinsic motivation
(relative to a no-competition/no-feedback/no-reward control group). The effect
size on intrinsic motivation was very large (see Deci, Koestner, & Ryan, 1999).
However, because the data were collapsed across winners and losers, it is unclear
whether the effects of competing for a reward had the same effect for winners as
for losers. In other words, it has not been determined whether the negative effect
on intrinsic motivation of a competitively contingent reward condition (relative to
a no-competition/no-feedback/no-reward control group) applies to both winners
and losers.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
278 Vansteenkiste and Deci
Winning Versus Losing a Reward
Because it has seemed clear that failing to get a reward could imply incompe-
tence and lack of control over outcomes and thus undermine motivation, the vast
majority of experiments examining the effects of rewards on intrinsic motivation
have allowed all participants in the experimental condition to receive the rewards.
This way, any negative effects could not be a function of failing to attain the target
goal. In all of the approximately 100 tangible-rewards experiments included in the
Deci et al. (1999) meta-analysis of tangible-reward effects on intrinsic motivation,
every participant in every reward condition received a tangible reward.
The Pritchard et al. (1977) study is the one reward study where not every
participant in the experimental group received a reward, and it also happened to
be the only study where the reward contingency involved direct competition (i.e.,
where the reward was competitively contingent). Thus, it is the only study that
included some participants who won the reward and some who lost it. As noted,
however, because of the way the data were presented, the effects on intrinsic
motivation of winning versus losing a competitively contingent reward were not
examined.
Accordingly, in the present study we explored the effects of winning versus
losing a competitively contingent monetary reward. While they were competing,
participants knew that winners would get a $3 reward but that losers would not,
and we examined the subsequent intrinsic motivation of winners relative to losers
of this reward. Presumably, because winning a competition when no rewards were
at stake led to higher intrinsic motivation than losing, winning a competitively
contingent reward should also lead to more subsequent intrinsic motivation than
losingthe reward.Thus, we predictedthatwinners woulddisplayhigher subsequent
intrinsic motivation than losers.
Many real-world situations in which people compete for rewards tend to be
relatively pressured—that is, beating the opponent and winning the reward tends
to be fairly strongly emphasized. Accordingly, to increase ecological validity, in
the competitive conditions of this study, the importance of winning was made quite
salient, so it was more pressured than the low-pressure competition condition of
the Reeve and Deci (1996) study but a bit less pressured than the high-pressure
condition of that study. Further, in the present study, we compared the subsequent
intrinsic motivation of those who won versus lost the competitively contingent
reward to that of participants in a no-competition/no-feedback/no-reward control
group similar to that used by Reeve and Deci. We expected that the difference in
the intrinsic motivation of winners versus losers of the competitively contingent
reward in this moderately pressured context would be primarily a function of
the losers having less intrinsic motivation than the control group rather than the
winners having more. That is, based on the results of the Reeve and Deci study,
we expected that the intrinsic motivation of winners in the relatively pressuring
context would not differ significantly from that of the control group.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 279
Intrinsic Motivation and Ego Involvement
When people are intrinsically motivated for an activity, they do it volitionally
andthey willingly persist at the activitywhen no external contingencies are present,
experiencing interest and enjoyment while doing so. People can, however, also be
motivated by other internal processes that lead them to persist at an activity even in
the absence of external contingencies. One such process is ego involvement (Ryan,
1982). When people’s egos—that is, their feelings of worth—are contingent on
some outcome such as doing well at a task, they may be very motivated to engage
that task, with a sense of resolve and determination, feeling pressure and tension
about having to do well. Ryan, Koestner, and Deci (1991) found that when people
became ego involved and then failed to do well at the ego-involving activity, they
tended to be persistent and insistent with respect to the activity—that is, they had a
high level of ego-involved motivation. Accordingly, Reeve and Deci (1996) found
that losers of a competition, although their intrinsic motivation for the task was
significantly undermined, did have a high level of ego-involved motivation for
the task. Apparently, the competition had gotten them ego involved in the task,
and losing made them determined to get better at the task in order to prove their
worth. As such, we expected that, in the current study, participants who lost the
competitively contingent rewards would show an enhanced level of ego-involved
motivation for the activity, needing to do better at in order to feel better about
themselves.
Is There Any Hope for Losers’ Intrinsic Motivation?
In the present study, we also examined possible ways of counteracting what
we expected would be strongly negative effects on intrinsic motivation of losing
the competitively contingent reward. We began by nothing that, with competitively
contingent rewards, losers get the negative performance feedback implicit in losing
and they also fail to get the desired rewards. Thus, we considered two possible
approaches to counteracting the negative effects of losing the competitively con-
tingent reward: (1) providing positive feedback on the quality of performance, and
(2) providing a performance-contingent reward to make up for the competitively
contingent reward they failed to win. Consider these in turn.
Positive Feedback for Losers
Positive performance feedback has generally been found to enhance intrinsic
motivation (Deci et al., 1999). Further, the initial study of the effects of competi-
tion on intrinsic motivation (Deci et al., 1981) indicated that when people received
implicit positive feedback, namely seeing that they solved puzzles faster than an-
other participant, they were significantly more intrinsically motivated than people

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
280 Vansteenkiste and Deci
who won a competition. Together these findings suggest that positive performance
feedback can be a powerfully positive motivator.
We therefore hypothesized that, if losers of a competitively contingent reward
received positive competence feedback about their actual performance at the activ-
ity,theywould be significantly more intrinsically motivatedthan comparable losers
who did not receive the positive feedback. In other words, the positive feedback
should help to counteract the negative feedback implicit in losing a competitively
contingent reward. In the present experiment, one of the groups of participants who
were told they could win a competitively contingent reward but in fact lost it were
also told at the beginning of the experiment that if they solved three out of the four
puzzles, they would be doing very well because they would have performed better
than 70% of previous participants. These participants, all of whom did in fact solve
at least three puzzles, thus got positive feedback about their performance, and we
expected them to show significantly higher intrinsic motivation than those who lost
the competitively contingent reward but got no positive performance feedback.
Monetary Rewards for Losers
As noted, losing a competitively contingent reward means not only that
participants get implicit negative feedback but also that they fail to attain their
goal of receiving the reward. As such, it is possible that, even though they lose
the competitively contingent reward, receiving a reward for doing well (i.e., a
performance-contingent reward) that is comparable in magnitude to the reward
theylost might counteract the negativeeffect of losing the competitivelycontingent
reward.
Tangible rewards, collapsed across type of contingency, have been found
meta-analytically to undermine both behavioral and self-report measures of in-
trinsic motivation (Deci et al., 1999). However, the meta-analysis showed that
performance-contingent rewards undermined the behavioral measure of intrinsic
motivation, but not self-reports of enjoyment. Our interpretation of this seemingly
disparate result is based in the idea that people enjoy receiving a performance-
contingent reward because it represents competence affirmation. Specifically, al-
thoughtangible performance-contingent rewards tend tobeexperienced as control-
ling and thus to decrease self-initiation of the activity, the message of competence
conveyed by the performance-contingent rewards promotes enjoyment. As such, a
performance-contingent reward should be nearly as effective as positive feedback
in terms of promoting enjoyment because in both conditions people will feel more
competent. However, if people are controlled by a performance-contingent reward,
they should not self-initiate the activity in the absence of an operative reward con-
tingency, because research has shown that perceived competence does not enhance
intrinsically motivated behavior if people do not feel autonomous (Fisher, 1978;
Ryan, 1982). Accordingly, we hypothesized that performance-contingent rewards

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 281
given to the losers of a competitively contingent reward would lead to less intrinsic
motivation behaviorally assessed than would positive feedback given to losers of
the competitively contingent reward, but that the levels of enjoyment for the two
groups would not differ.
Recently, although the evidence for this is relatively scant, Harackiewicz and
Sansone (2000) argued that a condition in which participants get performance-
contingent rewards for matching evaluative norms should yield higher levels of
intrinsicmotivation thana condition in which peopleare told their performance will
be evaluated and are then given positive feedback for matching evaluative norms.
(Those investigators made no distinction between the two measures of intrinsic
motivation.) In the current study, losers in the positive-feedback condition and the
performance-contingent rewards conditions were given evaluative norms before
they began the puzzle solving. Thus, counter to our predictions that performance-
contingent rewards (relative to positive feedback) would undermine free-choice
behavior but that the two groups would not differ on enjoyment, if Harackiewicz
and Sansone were correct, then the performance-contingent rewards condition
shouldhave ahigherlevel ofboth free-choice behavior andself-reported enjoyment
than would the positive feedback condition.
Summary and Hypotheses
The present study contained five experimental conditions. Condition 1 was a
no-competition/no-feedback/no-reward control group in which participants
worked alone on the target puzzle-solving activity used in all conditions. In the
other four conditions, participants competed against an apparent other participant
in the next room to try to win a $3 competitively contingent reward. In Condition 2,
the first of the four competitively contingent reward conditions, participants were
told that they won the competition and the $3 reward, whereas, in the other three
competitively contingent reward conditions, participants were told that they lost
the competition and the $3. In Condition 3, the first of the three losing conditions,
participants experienced only the loss of the competitively contingent reward. In
Condition 4, the second losing condition, participants lost the competitively con-
tingent reward but received positive competence feedback for solving at least three
out of the four puzzles, and thus performing better than 70% of previous partici-
pants. Finally, in Condition 5, the third losing condition, participants lost the $3
competitively contingent reward but they received a $3 performance-contingent
reward for solving at least three out of the four puzzles and thus performing better
than 70% of previous participants.
We made five primary predictions that were analyzed with contrasts and are
shown in Table I. Specifically, we hypothesized that the intrinsic motivation of
losers of the competitively contingent reward (with no positive feedback) would
be significantly lower than that of winners of the competitively contingent reward

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
282 Vansteenkiste and Deci
Table I. A priori Contrast Effects to be Tested
Losing c-c
Losing c-c reward,
No-competition, Winning Losing c-c reward, receiving receiving p-c
no-feedback, no-reward c-c reward reward positive feedback reward
control group (Condition 1) (Condition 2) (Condition 3) (Condition 4) (Condition 5)
Intrinsic motivation as the dependent variable
Losing c-c reward relative to winning 0 1 −100
c-c reward
Losing c-c reward relative to 1 0 −100
no-competition, no-feedback,
no-reward control group
Losing c-c reward with positive 0 0 1 −10
feedback relative to just losing
c-c reward
Losing c-c reward, but getting p-c 0 0 0 1 −1
reward relative to losing c-c reward,
but getting positive feedback
Ego-involved persistence as the dependent
variable
Losing c-c reward versus winning c-c 3 3 −2−2−2
reward and no-competition,
no-feedback, no-reward control group
Note. c-c reward =competitively contingent reward; p-c reward =performance-contingent reward.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 283
(Contrast1) and of theno-competition/no-feedback/no-reward controlgroup(Con-
trast 2). Next, we hypothesized that the intrinsic motivation of losers of the com-
petitively contingent reward who received positive performance feedback would
be significantly higher than that of losers with no positive feedback (Contrast 3).
Further, we hypothesized that the free-choice behavior of losers of the competi-
tively contingent reward who received performance-contingent rewards would be
significantly lower than that of the losers of the competitively contingent reward
who received positive performance feedback (Contrast 4), although the enjoyment
of the two groups was not expected to differ. Finally, we expected the ego-involved
motivationof the three groups of participants who lostthecompetitivelycontingent
rewardwouldbe higher than thatof the participantswho won thecompetitivelycon-
tingent reward and the participants in the no-competition/no-feedback/no-reward
control group who would not have become ego involved (Contrast 5).
Measuring Intrinsic Motivation and Ego-Involved Persistence
As already noted, past experiments have used two different measures of in-
trinsic motivation. The behavioral measure assesses behavior during a so-called
free-choice period subsequent to the performance period, and the self-report mea-
sure assesses participants’ reports of interest/enjoyment for the activity, which is
also done following the performance period. Consider the behavioral measure be-
cause it relates not only to intrinsic motivation but also to ego involvement that
can be easily stimulated in competitive situations.
The free-choice behavioral measure involved leaving participants alone with
the target activity and interesting alternative activities for a period of time when the
participantsassumedthatnoonewouldknow what they were doing and they would
notbeaskedto work more with the target activity (Deci, 1971). If participants spent
time with the target activity, they were assumed to be intrinsically motivated. The
number of seconds spent with the target activity was the measure of intrinsic
motivation.
This free-choice behavioral measure has been widely used, and it worked
extremely well to assess intrinsic motivation following an experimental manipu-
lation in which an external contingency was presented and subsequently removed.
However, it ran into problems when research began to investigate the effects of ego
involvement on intrinsic motivation (Ryan, 1982). Specifically, when the experi-
menter did a manipulation that resulted in people’s egos or feelings of self-worth
becoming contingent on their performance at an activity, there was no easy way for
the experimenter to remove the contingency. In other words, whereas an experi-
menter can make clear before the free-choice period begins that a tangible-reward
contingency is no longer operative, there is no comparable way to terminate an
ego-involvement contingency because it is within the person’s psyche. Thus, fol-
lowing an experimental period in which people became ego involved in their

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
284 Vansteenkiste and Deci
performance on a task, they may continue to work on the activity during a free-
choice period in order to buttress a threatened sense of worth. Thus, the free-
choice period behavior could reflect either intrinsic motivation or ego-involved
persistence.
Two approaches have been used for dealing with this problem. First, Ryan
et al. (1991) suggested that if free-choice behavior is really intrinsically moti-
vated, the person should be experiencing interest/enjoyment while doing it; in
other words, the free-choice behavior and self-reports of interest should be posi-
tively correlated. On the other hand, if the free-choice behavior is ego involved,
free-choice behavior and self-reported interest should not be correlated. Ryan
et al. further suggested that when ego-involved people do badly at a task, they
will be driven to persist during the free-choice period in order to regain a sense
of self-worth, whereas if they do well, they will not need to persist because their
good performance will have confirmed their worth. The researchers did an ex-
periment in which, in some conditions, any free-choice behavior was expected
to be intrinsically motivated and in others (viz., those that were ego-involving
where people did poorly) free-choice behavior was expected to be ego-involved
persistence. Results indicated that in the conditions where intrinsic motivation
was expected, free-choice behavior was significantly positively correlated with
expressed interest; whereas, in the conditions where ego-involved persistence was
expected, free-choice behavior was not correlated with expressed interest. As such,
Ryan et al. were able to separate the two types of free-choice behavior. However,
the disadvantage to this approach is that the free-choice period behavior was de-
fined as either intrinsic motivation or ego-involved persistence at the level of the
experimental condition rather than within individuals.
Thesecondapproach to distinguishing intrinsic motivationfrom ego-involved
persistence during a free-choice period was suggested by Reeve and Deci (1996).
They partitioned the free-choice measure into two categories: (a) working with
new puzzle configurations, and (b) working with the previously encountered con-
figurations. The reasoning was that, if people are intrinsically motivated for an
activity, they would seek novelty and challenge because of interest, so working
with new puzzles was considered an indicator of intrinsically motivated persis-
tence. In contrast, if participant are working to regain their feelings of worth—for
example, after having failed while ego involved—they would return tothe puzzles
on which they had done poorly in order to try to feel better about themselves. Thus,
working with old (previously-encountered) puzzles was considered an indicator
of ego-involved persistence.
Asalreadymentioned,competitiontendstopromoteegoinvolvement(Butler,
1989; Frederick & Ryan, 1995), and failing at a task in which one is ego involved
can promote ego-involved persistence. It was thus necessary to have measures of
bothintrinsic motivationandego-involvedpersistence in thecurrentstudy.Weused
free-choice time spent with new puzzles as the primary measure of intrinsic moti-
vation and we considered engagement with old puzzles a measure of ego-involved

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 285
persistence. As already noted, we hypothesized that ego-involved persistence (viz.,
time spent on old puzzles) would be greater in the lose conditions than in the win
and control conditions.
METHOD
Participants
A total of 25 male and 55 female University of Rochester undergraduates
volunteered for the experiment and received extra course credit for doing so. Most
of the participants were Caucasian in the age range of 18–23 years old.
Experimental Task
The experimental task was a spatial-relations puzzle called Happy Cubes.
Past studies with college students working on the Happy Cubes puzzle indicate
that participants find it to be a highly interesting activity and persist at it during
the free-choice period (e.g., Reeve et al., 1985; Reeve & Deci, 1996). Participants
were provided with drawings of three-dimensional configurations that they were
to reproduce with the mechanical puzzle.
Procedure
Participants arrived at a waiting area where they were welcomed by an ex-
perimenter and taken to the experimental room. The experimenter then went to an
adjoining room from where he observed participants through a one-way mirror and
communicated with them over an intercom. The experimenter explained that the
purpose of the study was to examine motivation for working on a spatial-relations
puzzle task. Participants were randomly placed in one of the five experimental
conditions. Those in the four competitive conditions were told that their same-sex
opponent had already arrived and was settled in the adjoining room. The experi-
menter used a sex-appropriate name to refer to that “other participant.” In reality,
there was no person in the next room, but the purpose of this pretext was to help
create a competitive atmosphere. On the table in all five conditions were (a) the
Happy Cubes puzzle, (b) two drawings of three-dimensional configurations that
the participants would attempt to duplicate with the puzzle during a practice pe-
riod, and (c) a selection of popular magazines that served as an alternative activity
during a later free-choice period. The experimenter explained that there would be
a series of six puzzles to solve. The first two of which were there on the table and
would be the practice puzzles. The other four would be used subsequently as the
actual performance puzzles.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
286 Vansteenkiste and Deci
Before starting the practice puzzles, participants received experimental in-
structions appropriate to their experimental condition. In the no-competition/no-
feedback/no-reward control group (Condition 1) the experimenter simply asked
the participants to work on the puzzles, “doing your individual best.” In each
of the four competitively contingent reward conditions, the experimenter told
participants the following: “The purpose of this competition is to try to outperform
the other person by solving your puzzles faster than he or she.” Thus, participants
were led to believe that they would to be competing against the same-sex partic-
ipant who had been said to be in the adjacent room. “To win the game you will
need to solve the puzzles more quickly than your opponent. You will have four
puzzles to solve and you will have four minutes to solve each puzzle. It does not
matter whether you figure out how the puzzle works, the important thing is to
win the competition. So, focus your attention on being the winner.” In all four
competitively contingent reward conditions (Conditions 2–5), participants were
told that “you can make $3 by winning the competitive game; in other words, you
will get $3 if you solve more of the puzzles more quickly than your opponent.”
People in Conditions 4 and 5 were given the following additional instructions: “if
you manage to solve three out of the four puzzles correctly, regardless of whether
you are slower or faster than your opponent, you will have done better than 70%
of the participants in our previous studies. So, the purpose of this competition is
twofold: first, to try to beat the other person by solving your puzzles faster than
he (or she) does, and second, to try to surpass the 70th percentile performance
standard by correctly solving at least three of the four puzzles within the allotted
time.” People in Condition 5 were then given an additional instruction, namely
that “if you meet the 70th percentile performance standard by correctly solving at
least three out of the four puzzles, you will receive a $3 reward. So, there are two
ways to earn a reward of $3; one way is through beating the other person and the
other is by performing well, namely by solving three out of four puzzles correctly.
If you solve three puzzles and win the game you will earn $6.” In order to ensure
that all participants in Conditions 4 and 5 would be able to solve at least three of
the four puzzles we did extensive pretesting. Our aim was to find four puzzles that
would be as challenging as possible but would allow all participants to solve at
least three of the four.
The amount of money given for the performance-contingent reward was $3
just as the amount given for the competitively contingent reward was $3. This was
done so the losers of the competitively contingent reward in Condition 5 would
receive the same amount of money from the performance-contingent reward that
the winners of the competitively contingent reward received in Condition 2, and
also so the losers in Condition 5 would lose the same amount as the losers in
Conditions 3 and 4 who ended up without any monetary reward. The losers in
Condition 5 would have imagined that the winners had made three dollars more
than they themselves received (because they would have believed that the winner

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 287
would get $3 for winning and $3 for solving at least three puzzles). Similarly,
the losers in Conditions 3 and 4 would have imagined that the winners had made
$3 more than they did from winning the competition (these participants would
not have known anything about a performance-contingent reward). As such, for
each comparison, the amount of money believed to have been won and/or lost was
controlled across relevant comparisons.
Before starting the performance puzzles and after receiving the instructions,
participantsin all conditionsweregiven twopractice puzzles. Afterthefirst, partici-
pantsin the four competitivelycontingent reward conditions(Conditions2–5) were
informedthat their opponenthad failed tosolvethe puzzlein the allotted time,so the
participant would have won on that puzzle. After the second practice puzzle, par-
ticipants were told that their opponent had solved the second puzzle faster than they
did. Puzzles had been preselected so that the participants would not be able to solve
the second practice puzzle within the allotted time, but could solve the first practice
puzzle. This matching of practice outcomes in the competition conditions was in-
tended to convey the impression that the two people were roughly of equal ability
and was also intended to create the perception of a challenging game. After the
two practice trials the experimenter told participants the practice period was over.
The experimenter then entered the room to remove the practice configura-
tions and leave six more configurations. Four of these configurations were for the
performance phase of the experiment and were put on one end of the table. The
other two were placed on the other end of the table so participants could work
on new puzzles during the subsequent free-choice period. The period of working
on the performance puzzles then began. In each of the competitively contingent
reward conditions, the experimenter provided comparison feedback after each trial
by telling the participants over the intercom which of them (i.e., the participant
or the “confederate” opponent) had won on that puzzle. The feedback that was
provided had been predetermined according to condition. The win versus lose
outcome was manipulated by the experimenter as follows. The score of the com-
petitive set of four performance puzzles was always 3-1. Participants in the win
condition (Conditions 2) were told they won the first, third, and fourth trial and the
opponent had won the second trial. Participants in the lose conditions (Conditions
3–5) were told they lost the first, third, and fourth trial, but they had won the sec-
ond trial. This way, the score after two trials was equal (1-1) in all competitively
contingent reward conditions to further the feelings of competition and challenge.
If participants won a puzzle, they had to wait an additional minute or so, in order
to give the impression that their opponent was still working on the puzzle.
After the four performance puzzles were completed, participants in all com-
petitive conditions were given the results of the competition, and those in Con-
ditions 4 and 5 got additional feedback. Specifically, participants in Condition 2
were informed that they had won the competition and would receive their $3 for
winning at the end of the experiment. Participants in Conditions 3, 4, and 5 were

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
288 Vansteenkiste and Deci
informedthatthey had lost the competition and wouldnotreceive the competitively
contingent reward. In Conditions 4 and 5, participants were told that, “although
you lost the competition, you still managed to solve at least three out of the four
puzzles correctly. I can tell you that you did very well because you reached the
performance standard of doing better than 70% of the participants in previous
experiments.” For participants in Condition 5, the experimenter also said, “So, at
the end of the experiment you will receive the $3 reward I promised you.” Because
the four performance puzzles had been pretested all participants were able to solve
at least three of the puzzles within the allotted time, so all competitive losers in
Condition 4 received the positive feedback and all competitive losers in Condition
5 receive the performance-contingent reward.
After the four performance trials, the experimenter entered the experimental
room and told participants that the puzzle-solving phase of the experiment was
over and that he needed to go to another room to get a final questionnaire for them.
Because the appropriate questionnaire was to be prepared by the computer based
on their performance, it would take him just a few minutes to get the two correct
questionnaires and drop off the one for the opponent. The participants were left
alone for a 6-min free-choice period, during which they could do whatever they
wanted to do in the experimental room until the experimenter returned. During this
period an observer recorded the amount of time the participants spent on the new
and the old puzzle configurations. The observers had been trained to be able to
look through the one-way window and identify whether particular puzzles were the
performance-period puzzles (i.e., old puzzles) or the free-choice-period puzzles
(i.e., new puzzles). After the free-choice period, the experimenter reentered the
room, administered the postexperimental questionnaire, debriefed the participants,
and gave the reward to those who had been told they would receive it.
Behavioral Measure
The behavioral measure of intrinsic motivation was taken during the 6-min
free-choice period subsequent to the feedback that was provided at the end of the
performance period. Specifically, the amount of time participants spent working
with new puzzles during this free-choice period was recorded by the observer who
was blind to the hypotheses and experimental conditions. The amount of free-
choice time spent on previously-encountered (i.e., old) configurations was also
recorded as the measure of ego-involved persistence.
Questionnaire Measures
Participants completed a questionnaire after the free-choice period. This pos-
texperimental questionnaire included 13 items (scored on 7-point scales), 7 items

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 289
measuring “enjoyment of the activity” (e.g., “I enjoyed doing this puzzle-solving
task very much.”; α=.90) which was used as the self-report measure of intrin-
sic motivation; and 6 items measuring “perceived competence” (e.g., “I think I’m
pretty good at this activity”; α=.93) which was used as a manipulation check
to ensure that winners felt more competent than losers and that losers who got
positive feedback felt more competent than losers who did not. Also, because past
studies have shown that preceived competence mediated the effects on intrinsic
motivation of winning versus losing a competition and of positive feedback, its
inclusion allowed us to test whether it would mediate any condition effects in this
study.
RESULTS
Preliminary Analyses
We began by examining the effectiveness of the two approaches to differentia-
ting intrinsic motivation from ego-involved persistence, namely, the within-person
and the within-group approaches. We used as a criterion that free-choice be-
havior theorized to reflect intrinsic motivation would be correlated with enjoy-
ment whereas free-choice behavior theorized to reflect ego-involved persistence
would not. First, we considered the within-person approach by correlating self-
reports of enjoyment with the two types of free-choice behavior—working with
new puzzles versus working with previously-encountered puzzles. As in Reeve
and Deci (1996), a significant positive correlation was found between working
on new puzzles and enjoyment, r(78) =.23, p<.05, but working on old puz-
zles was not correlated with enjoyment, r(78) =.05, p<.64, thus suggesting
that time spent working on new puzzles is a good index of intrinsic motivation
(see Table II).
Second, we examined the group-level approach by correlating total free-time
behavior (the sum of the time spent working on new puzzles and old puzzles) with
self-reported enjoyment in Conditions 1 and 2 versus Conditions 3–5 to examine
whether total free-choice behavior would tend to be more reflective of intrinsic mo-
tivation in the former two conditions and of ego-involved persistence in the latter
Table II. Correlations Among All Variables
123 45
1. Perceived competence —
2. Enjoyment .56∗∗ —
3. Time spent on new puzzles .17 .23∗—
4. Time spent on old puzzles −.03 .05 −.18 —
5. Total time spent on puzzles .15 .25∗.88∗∗ .32∗∗ —
∗p<.05. ∗∗ p<.01.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
290 Vansteenkiste and Deci
three. Ryan et al. (1991) argued that ego-involving conditions (e.g., competition)
where people get negative feedback are the most likely to lead to ego-involved
persistence during the free-choice period. Accordingly, in conditions where there
was implicit negative feedback from losing the competition (Conditions 3–5), to-
tal free-choice behavior and enjoyment were not correlated, r(46) =.16, p<.40,
whereas, in conditions with no competition or no implicit negative feedback (Con-
ditions 1 and 2), the magnitude of correlation between total free-choice behavior
and enjoyment was more than twice as large and nearly significant, r(30) =.34,
p<.06. This is the first study in which the within-person and the group-level
approaches were both considered, and these preliminary results suggest that both
approaches are reasonable ways to distinguish the two types of persistence.
In the next preliminary analysis, we examined perceived competence as a ma-
nipulation check using protected ttests. Competitors in Condition 2 who won the
reward felt more competent than losers in Condition 3 who got no feedback and no
reward, t(75) =4.09, p<.001. Losers in Condition 4 who received positive per-
formance feedback felt significantly more competent than loser in Condition 3
who received neither feedback nor rewards, t(75) =−2.95, p<.01. Finally,
losers in Condition 5 who got performance-contingent rewards did not differ in
their level of perceived competence from losers in Condition 4 who got positive
performance feedback, t(75) =−0.32, p<.75. As such, the manipulations did
work in that losing diminished feelings of competence whereas positive feedback
and performance-contingent rewards enhanced feelings of competence.
Primary Analyses
A MANOVA was conducted across the five conditions with all five behavioral
and self-report variables included. The overall F-value for the Pillai’s procedure
was significant, F(16,300) =2.49, p<.001. Subsequently a one-way ANOVA
was performed for each of the variables, and then the hypothesized contrasts were
tested. The cell-means appear in Table III.
The five experimental conditions differed significantly on the intrinsic moti-
vation behavioral measure of working on new puzzles, F(4,75) =4.48, p<.01,
and on perceived competence, F(4,75) =4.89, p<.001. The conditions were
marginally different on self-reported enjoyment, F(4,75) =2.01, p<.10, on
ego-involved persistence (i.e., playing with old puzzles), F(4,75) =2.37, p<
.06, and on total free-choice behavior (i.e., the sum of working on old and new
puzzles), F(4,75) =2.02, p<.10.
Contrast Analyses
Next, we performed contrasts using protected t tests to examine the hy-
potheses about intrinsic motivation and ego-involved (i.e., internally-controlled)

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 291
Table III. Means and Standard Deviations for Perceived Competence, Enjoyment, and Number of Seconds Spent on the New or/and Old Puzzles
Losing c-c
reward, Losing c-c
No-competition, receiving reward,
no-feedback, Winning Losing c-c positive receiving p-c
no-reward c-c reward reward feedback reward
(Condition 1) (Condition 2) (Condition 3) (Condition 4) (Condition 5)
Mean SD Mean SD Mean SD Mean SD Mean SD
Perceived competencea4.88 0.99 5.27 1.01 3.66 1.34 4.82 1.11 4.94 1.06
Enjoymenta5.31 0.98 5.34 0.78 4.62 1.14 5.42 0.85 5.15 0.74
Time spent on new puzzles 205.1 144.1 217.4 132.9 95.0 132.1 199.6 153.9 66.4 95.1
Time spent on old puzzles 18.3 45.8 6.6 13.8 27.5 42.2 33.8 75.1 78.3 124.7
Total time spent on puzzles 223.4 151.77 224.0 133.7 122.5 148.1 233.4 154.2 143.5 140.2
Note. c-c reward =competitively contingent reward; p-c reward =performance-contingent reward. n=16 per cell.
aResponse rates varied between 1 (Completely Disagree)and7(Completely Agree).

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
292 Vansteenkiste and Deci
persistence. First, as predicted for Contrast 1, participants who lost the competition
andgotneitherpositivefeedback nor rewards (Condition 3) spent less time with the
newpuzzles duringthefree-choice period, t(75) =2.60, p<.01,and reportedless
enjoyment, t(75) =2.28, p<.03, than did those who won the competition and
got the contingent reward (Condition 2). Further, as predicted for Contrast 2, losers
of the competition were less intrinsically motivated on both the free-choice mea-
sure, t(75) =2.34, p<.02, and the self-report measure, t(75) =2.14, p<.04,
than were participants in the no-competition/no-feedback/no-reward control group
(Condition 1).
Now consider the three groups who lost the competitively contingent re-
ward. As predicted for Contrast 3, those who received specific positive feedback
for performing above the 70th percentile (Condition 4) were significantly more
intrinsically motivated than losers who got no positive feedback (Condition 3),
as reflected in both the free-choice measure, t(75) =−2.22, p<.05, and the
self-report measure, t(75) =−2.50, p<.01. Further, as expected for Contrast
4, providing performance-contingent rewards to losers for performing above the
70th percentile (Condition 5) resulted in significantly less free-choice behavior
on new puzzles than providing the comparable positive feedback without rewards,
t(75) =2.95, p<.01, but there was not a difference between the two groups in re-
ported enjoyment, t(75) =0.94, p<.35. Parenthetically, although these contrasts
were not specified in the hypotheses, losers who received performance-contingent
rewards did not differ in their level of free-choice behavior from losers who got nei-
ther feedback nor rewards, t(75) =0.64, p<.54, but the performance-contingent
rewards group was marginally higher on enjoyment, t(75) =−1.69, p<.09.
Concerning ego-involved persistence, our one prediction for Contrast 5 was
that the three groups of losers would be higher in ego-involved persistence (i.e.,
time spent on the old puzzles) than would winners and control group (i.e., non-
ego-involved) participants, and the contrast supported this hypothesis, t(75) =
−2.10, p<.05. This finding supplemented the preliminary, group-level analysis
that showed that total free-choice behavior in the lose conditions tended to reflect
more ego-involved persistence whereas in the win and control conditions it tended
to reflect more intrinsic motivation.
Mediational Analyses
We assessed perceived competence primarily as a manipulation check; how-
ever, because it varied by condition as reported above and because it has mediated
the effects on intrinsic motivation of winning versus losing a competition in pre-
vious studies (Vallerand & Reid, 1984), we examined whether it would mediate
the intrinsic motivation contrast effects in this study. As a preliminary analysis,
we tested whether actual performance varied by condition in order to ensure that
any mediation by perceived competence of the contrast effects would not be a

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 293
function of actual puzzle performance. We performed a one-way ANOVA on the
five conditions for the average time taken by participants to solve the four puzzles
and found that the result was not significant, F(4,75) =0.77, p<.55.
To examinethe degree to which perceivedcompetencemediated the condition
effects on intrinsic motivation, we used the regression procedure suggested by
Judd and Kenney (1981), first for intrinsic motivation assessed as new-puzzle
behavior and then for intrinsic motivation assessed as enjoyment. Mediation can
be concluded if a significant effect of the independent variable on the dependent
variable decreases in magnitude and becomes nonsignificant when the mediator
(perceived competence) is added to the equation, assuming the mediator remains
a significant predictor of the outcome.
Consider new puzzle behavior. All four hypothesized effects (namely, win
versus lose; control group versus lose; lose versus lose plus positive feedback;
and lose plus positive feedback versus lose plus performance-contingent reward)
which were tested with Contrasts 1–4 for new puzzle behavior were significant,
as outlined above. For each, then, the first requirement of the Judd and Kenny
procedure was satisfied—the independent variable was significantly related to the
dependent variable. However, as can be seen in the correlation matrix of Table II,
perceived competence was not significantly correlated with new-puzzle behavior,
so the requirement specified by Judd and Kenny that the mediator be related to
the dependent variable was not met. Accordingly, we concluded that perceived
competence did not mediate any of the contrast effects on the free-choice measure
of intrinsic motivation.
Now consider the dependent variable of self-reported enjoyment. As pre-
dicted, only three of the four contrast effects were significant in the ANOVAs—
namely, win versus lose; control group versus lose; and lose versus lose plus
positive feedback. Thus, the first requirement for mediation was met for these
three relations. Thus, we examined possible mediation by perceived competence
for each of these relations.
We began by examining the effect of winning versus losing (with no addi-
tional feedback or reward). The contrast effect did predict perceived competence,
β=.57, F(1,30) =14.67, p<.001, and perceived competence predicted en-
joyment when the effect of condition was controlled, β=.54, F(2,29) =8.33,
p<.01, so these requirements for mediation were met—namely, the indepen-
dent variable predicted the mediator and the mediator predicted the dependent
variable when the independent variable was controlled for. Finally, the significant
effect of the competitive-outcome contrast became nonsignificant as a predictor
of enjoyment when perceived competence was added to the equation, β=.06,
F(2,29) =0.08, p<.77. As such, the analyses suggest that the effect on enjoy-
ment of winning versus losing a competitively contingent reward was mediated by
perceived competence.
Then, we turned to whether the undermining of enjoyment by losing relative
to the control group was mediated by perceived competence. In this analysis, the

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
294 Vansteenkiste and Deci
contrastfor the manipulated independent variablepredicted perceived competence,
β=.47, F(1,30) =8.52, p<.01, and the relation of perceived competence to
enjoyment when controlling for the effect of the contrast was also significant,
β=.40, F(2,29) =5.24, p<.05, so these requirements for mediation were
met. Finally, the contrast effect of losing relative to the control group became
nonsignificant as a predictor of enjoyment when perceived competence was added
to the equation, β=.24, F(2,29) =1.83, p<.19. Thus, the analyses suggest
that the effect on enjoyment of losing a competitively contingent reward relative
to a no-competition/no-feedback/no-reward control group was also mediated by
perceived competence.
Last,we tested for mediation of the effecton self-reported enjoyment of losing
versus losing but receiving positive feedback. The independent variable did predict
perceived competence, β=.44, F(1,30) =7.07, p<.05, and the relation of
perceived competence to enjoyment when controlling for the effect of the contrast
was also significant, β=.43, F(2,29) =6.06, p<.05, so these requirements
for mediation were met. Finally, the contrast effect of losing relative to losing
but getting positive feedback became nonsignificant as a predictor of enjoyment
when perceived competence was added to the equation, β=.19, F(2,29) =1.20,
p<.28. As such, the analyses suggest that the effect of losing a competitively
contingent reward relative to losing the reward but getting positive feedback was
mediated by perceived competence.
DISCUSSION
In the typology of reward contingencies (Deci et al., 1999; Ryan et al., 1983),
the competitive contingency, which means receiving a reward for beating an op-
ponent, has been the least studied. The fact that it requires winning makes the
contingency quite controlling, and the one published study of people actually
competing for rewards showed the contingency to be highly undermining of in-
trinsicmotivationusing both the free-choicebehavioralmeasure and the self-report
measure (Pritchard et al., 1977). However, it is an interesting contingency because
participants who win the reward receive salient competence information. Presum-
ably, for winners, the feedback implicit in the competitive outcome should be
very affirming of competence, whereas for losers, it could be quite discouraging,
resulting in significantly different levels of intrinsic motivation as was the case
with winning versus losing a competition without contingent rewards (Reeve &
Deci, 1996; Vallerand & Reid, 1984). Thus far, no experiment had examined the
effects of winning versus losing competitively contingent rewards, so the first im-
portant finding of the present study was that winners of competitively contingent
rewards were significantly more intrinsically motivated than loser, assessed with
both free-choice behavior and self-reports of enjoyment. As well, winners felt
more competent than losers.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 295
The relations of the intrinsic motivation of a control group to that of the
winners and of the losers of a reward depend on what type of control group is
used, an issue that is very important in terms of one’s conclusions about the ef-
fects of competition and competitively contingent rewards. In the first study of
competition and intrinsic motivation, winners of a competition were less intrin-
sically motivated that the control group that did not compete but got positive
feedback comparable to that which was implicit in winning the competition (Deci
et al., 1981). In a later study, Reeve and Deci (1996) used a control group that
did not compete and did not get performance feedback. In that study, winners in
a nonpressuring interpersonal ambience were more intrinsically motivated than
participants in the no-competition/no-feedback control group, whereas winners
in a pressuring interpersonal ambience did not differ in intrinsic motivation from
the no-competition/no-feedback control group. Losers, on the other hand, showed
substantially less intrinsic motivation than the no-competition/no-feedback con-
trol group. Unlike these two previous studies, the present study examined not just
competition but winning versus losing a competitively contingent reward. In the
present study, which involved competing for a reward in a relatively pressuring
interpersonal ambience, we expected and found no difference between the intrin-
sic motivation of winners and participants in the no-competition/no-feedback/no-
reward control group, but losers were significantly less intrinsically motivated than
these control-group participants. Together, this set of findings suggests that get-
ting positive feedback in the presence of another doing the same activity is more
facilitative of intrinsic motivation than winning a competition against that other.
Further, losing a competition or a competitively contingent reward leads to less
intrinsic motivation than winning a comparable competition or competitively con-
tingent reward and also to less intrinsic motivation than a control group that does
not compete and either does or does not get positive feedback.
Perhapsthemostimportantaspectsofthecurrentexperimentwas the focus on
losers. Most previous studies of competition have included only winners, with the
exception of a few studies that compared winners versus losers of a competition,
andno study of competitivelycontingent rewardshas examined the effectsof losing
the reward. In the present study, expecting that losing a competitively contingent
reward would be quite detrimental to intrinsic motivation (which in fact it was), we
examined whether there was any way to ameliorate that negative effect. Because
losersofacompetitivelycontingent reward both got implicit negativefeedback and
lost the reward, we provided one group of losers with positive feedback for meet-
ing a specified standard and another group with performance-contingent rewards
for meeting the same specified standard (even though both groups lost the com-
petition). Losers who got neither positive feedback nor performance-contingent
rewards served as a comparison group for examining these questions.
Results of the present study revealed that positive performance feedback
can indeed go a long way toward counteracting the negative effects of losing.
Participants who lost the competition but got positive performance feedback were

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
296 Vansteenkiste and Deci
significantly more intrinsically motivated than losers who did not get the positive
feedback, and the losers with positive feedback did not differ significantly in
intrinsicmotivationfromeither the no-competition/no-feedback/no-rewardcontrol
group or the winners. This set of findings appeared for both the behavioral and
self-report measures. It seems therefore that positive feedback about performance
is extremely important for maintaining intrinsic motivation in competitive settings.
Indeed, as noted, Deci et al. (1981) found that receiving positive implicit feedback
in the presence of another doing the same task led to significantly greater intrinsic
motivation than winning the competition. In other words, it is positive information
rather than winning that is most nourishing of intrinsic motivation (McAuley &
Tammen, 1989).
The effects of performance-contingent rewards for losers were more complex
and help to sort out some of the confusion in the literature about performance-
contingent rewards. The most comprehensive and valid meta-analysis of reward
effects showed that performance-contingent rewards undermined the free-choice
behavioral measure of intrinsic motivation but did not undermine enjoyment of
the activity (Deci et al., 1999). Accordingly, we expected different results on the
behavioral versus self-report measure for losers who got performance-contingent
rewards. However, in making predictions about performance-contingent-reward
effects it is necessary to consider what type of control group is being used. There
are three possibilities:(a) a group that gets no rewards and no feedback; (b) a
group that gets no rewards but gets unexpected positive feedback comparable
to that implicit in the rewards for the experimental group; and (c) a group of
participantswho get no rewardsbut are toldthattheir performance will be evaluated
and subsequently are given positive feedback comparable to that implicit in the
rewards for the experimental group. The Deci et al. meta-analysis showed that
performance-contingent rewards undermined free-choice behavior relative to the
first two of these control groups. The third group was not included in the meta-
analysis because relatively few studies have been done using a control group
where participants were told their performance would be evaluated and then were
given positive feedback. Counter to our predictions that performance-contingent
rewards would be undermining relative to this evaluation-and-positive-feedback
control group, Harackiewicz and Sansone (2000) argued that relative to this control
group, performance-contingent rewards would enhance intrinsic motivation.
The current study used exactly that control group. The group of losers who
got the positive feedback were told before they began that if they solved three
out of the four puzzles correctly they would have performed well, better than
70% of previous participants, and subsequently they were told that they had done
well by meeting the standard. Results of the study showed that with the free-
choice behavioral measure, performance-contingent rewards undermined intrinsic
motivation relative to the group of participants who were told they would be
evaluated and then got positive feedback, whereas with the enjoyment measure

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 297
performance-contingent rewards left intrinsic motivation unchanged relative to the
group that got positive feedback for meeting the standard. Thus, the results were
in line with the Deci et al. meta-analysis and suggest that performance-contingent
rewards tend to undermine intrinsically motivated behavior but leave enjoyment
unchanged relative to all three types of control groups outlined above. As such,
these results, which supported our hypothesis, failed to support the Harackiewicz
and Sansone prediction that performance-contingent rewards would lead to greater
intrinsic motivation than this evaluation-and-positive-feedback control group.
There are two points worth making with regard to these results. First, the
majority of studies of performance-contingent rewards and of competition done
by the Harackiewicz and Sansone group have used only the self-report measure
that has tended not to show significant undermining of enjoyment by performance-
contingentrewards.It appears that the self-report measureisresponsiveto any indi-
catorof competence evenwhen the perceivedcompetence does not positivelyaffect
the motivation of behavior. Indeed, in the present study, performance-contingent
rewards did not undermine enjoyment relative to positive feedback (although it
did undermine free-choice behavior). Further, the performance-contingent rewards
group had marginally higher enjoyment than the group of losers who got neither
feedback nor rewards (although the free-choice behavior of those two groups did
not differ). The other point worth noting about these findings is that all participants
in both the positive feedback and the performance-contingent rewards groups had
lost the competition for rewards, and it is unclear how that might have affected the
relative intrinsic motivation of these two groups. Additional research comparing
performance-contingent rewards to an evaluation-and-positive feedback control
group using participants who had not previously lost a competitively contingent
reward would be important before definitive conclusions can be drawn. However,
such work has to include both a behavioral and self-report measure of intrinsic
motivation if it is to be useful, and it must be done in a way that ego-involved per-
sistence can be differentiated from intrinsic motivation, as was done in the current
study.
In this regard, the current study showed that people who competed and did
poorly (i.e., lost) persisted longer on old puzzles during the free-choice period
than did those who either did not compete or won. Further, for the participants
who lost, their total persistence did not relate to enjoyment, further suggesting that
their persistence was ego involved. This result is in line with the findings of Ryan
et al. (1991) who found such persistence with people who were induced to be ego
involved without explicit competition and then performed relatively badly.
One important conclusion that is emerging from the studies of competition
and competitively contingent rewards is that positive performance feedback is
crucial for maintaining intrinsic motivation in competitive settings. It is more
facilitative than winning a competition (Deci et al., 1981), and it can counteract
the negative effects of losing a competition or a competitively contingent reward.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
298 Vansteenkiste and Deci
In a sense, this suggests that the important thing about a competitive outcome is
how people feel about it—how they personally experience it. In fact, a study by
McAuley and Tammen (1989) showed that winning versus losing a competition
on a familiar activity did not have a significant effect on intrinsic motivation but
that people’s self-evaluations of how well they did on the task was significantly
related to intrinsic motivation. As such, this, along with the finding that winning
when feeling pressured to do so has a substantially negative effect on intrinsic
motivation indicates that a strong focus on winning, whether by parents, coaches,
or athletes, can be quite problematic for people’s intrinsic motivation.
It is important to note that the current studies and most of those reviewed
in this paper were conducted in psychology laboratories. Clearly, although this
allows for the degree of careful control necessary to disentangle what elements in
the competitive situation have what effects on intrinsic motivation, it does create
a problem for ecological validity. For example, the target activities used in lab
experiments are typically novel, whereas in the real world people often compete
at activities they have been training at for years and they may have a deep personal
commitment to doing well at the activity. In other words, the activities of life at
which people complete may be an important part of their identities, and there is
little evidence of how the results of laboratory experiments would generalize to
such real-world situation.
In conclusion, it seems that trying to win competitions and competitively
contingent rewards is becoming more and more prevalent in modern culture, yet
it appears that a focus on winning may indeed be counter-productive at least with
respect to intrinsic motivation for the target activities. If, instead of emphasizing
winning above all else, participants in activities and observers of the activities
focusedmore on goodperformancethan on winning,theresults for theparticipants’
motivation is likely to be far more positive.
REFERENCES
Butler, R. (1989). Interest in the task and interest in peers’ work in competitive and noncompetitive
conditions: A developmental study. Child Development,60, 562–570.
Deci,E. L.(1971). Effects ofexternally mediated rewardson intrinsicmotivation.Journal of Personality
and Social Psychology,18, 105–115.
Deci, E. L., Betley, G., Kahle, J., Abrams, L., & Porac, J. (1981). When trying to win: Competition
and intrinsic motivation. Personality and Social Psychology Bulletin,7, 79–83.
Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review of experiments examining
the effects of extrinsic rewards on intrinsic motivation. Psychological Bulletin,125, 627–668.
Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Human needs and the
self-determination of behavior. Psychological Inquiry,11, 227–268.
Epstein, J. A., & Harackiewicz, J. M. (1992). Winning is not enough: The effects of competition
and achievement orientation on intrinsic interest. Personality and Social Psychology Bulletin,18,
128–138.
Fisher, C. D. (1978). The effects of personal control, competence, and extrinsic reward systems on
intrinsic motivation. Organizational Behavior and Human Performance,21, 273–288.

P1: GCE
Motivation and Emotion [me] PP1017-474719-01 October 17, 2003 12:10 Style file version Nov 28th, 2002
Can Losers Remain Motivated? 299
Frederick,C. M., &Ryan, R. M.(1995). Self-determination insport: Areviewusingcognitiveevaluation
theory. International Journal of Sport Psychology,26, 5–23.
Harackiewicz, J. M., & Sansone, C. (2000). Rewarding competence: The importance of goals in the
study of intrinsic motivation. In C. Sansone & J. M. Harackiewicz (Eds.), Intrinsic and extrinsic
motivation: The search for optimal motivation and performance (pp. 79–103). San Diego, CA:
Academic Press.
Judd, C. M., & Kenney, D. (1981). Estimating the effects of social interventions. Cambridge, UK:
Cambridge University Press.
McAuley, E., & Tammen, V. V. (1989). The effects of subjective and objective competitive outcomes
on intrinsic motivation. Journal of Sport and Exercise Psychology,11, 84–93.
Pritchard, R. D., Campbell, K. M., & Campbell, D. J. (1977). Effects of extrinsic financial rewards on
intrinsic motivation. Journal of Applied Psychology,62, 9–15.
Reeve,J.,& Deci, E. L. (1996).Elements within the competitive situation thataffect intrinsic motivation.
Personality and Social Psychology Bulletin,22, 24–33.
Reeve, J., Olson, B. C., & Cole, S. G. (1985). Motivation and performance: Two consequences of
winning and losing a competition. Motivation and Emotion,9, 291–298.
Ryan, R. M. (1982). Control and information in the intrapersonal sphere: An extension of cognitive
evaluation theory. Journal of Personality and Social Psychology,43, 450–461.
Ryan,R. M.,Koestner,R., & Deci, E. L. (1991). Variedforms of persistence: When free-choice behavior
is not intrinsically motivated. Motivation and Emotion,15, 185–205.
Ryan, R. M., Mims, V., & Koestner, R. (1983). Relation of reward contingency and interpersonal
context to intrinsic motivation: A review and test using cognitive evaluation theory. Journal of
Personality and Social Psychology,45, 736–750.
Tauer, J. M., & Harackiewicz, J. M. (1999). Winning isn’t enough: Competition, achievement orienta-
tion, and intrinsic motivation. Journal of Experimental Social Psychology,35, 209–235.
Vallerand, R. J., Gauvin, L. I., & Halliwell, W. R. (1986). Negative effects of competition on children’s
intrinsic motivation. Journal of Social Psychology,126, 649–657.
Vallerand, R. J., & Reid, G. (1984). On the causal effects of perceived competence on intrinsic moti-
vation: A test of cognitive evaluation theory. Journal of Sport Psychology,6, 94–102.