Content uploaded by Ilona de Hooge
Author content
All content in this area was uploaded by Ilona de Hooge
Content may be subject to copyright.
This article was downloaded by:[Universiteit Van Tilburg]
On: 31 July 2007
Access Details: [subscription number 769141684]
Publisher: Psychology Press
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Cognition & Emotion
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713682755
Moral sentiments and cooperation: Differential
influences of shame and guilt
Online Publication Date: 01 August 2007
To cite this Article: de Hooge, Ilona E., Zeelenberg, Marcel and Breugelmans,
Seger M. (2007) 'Moral sentiments and cooperation: Differential influences of shame
and guilt', Cognition & Emotion, 21:5, 1025 - 1042
To link to this article: DOI: 10.1080/02699930600980874
URL: http://dx.doi.org/10.1080/02699930600980874
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly
forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents will be
complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be
independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or
arising out of the use of this material.
© Taylor and Francis 2007
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
Moral sentiments and cooperation: Differential
influences of shame and guilt
Ilona E. de Hooge, Marcel Zeelenberg, and
Seger M. Breugelmans
Tilburg University, Tilburg, The Netherlands
For centuries economists and psychologists (Frank, 1988; Ketelaar, 2004; Smith,
1759) have argued that moral emotions motivate cooperation. Ketelaar and Au
(2003) recently found first evidence that guilt increases cooperation for proselfs in
social bargaining games. We investigated whether this effect would also occur for
shame, another moral emotion. Using a dyadic social dilemma game in Experiment
1 and an everyday cooperation measure in Experiment 2 as measures for short-term
cooperation, we replicated Ketelaar and Au’s findings for guilt. However, as
predicted on the basis of previous emotion research, we found no such effect for
shame. These results clearly indicate that the effects of moral emotions on
cooperative behaviour can only be understood if the specific moral emotion is
known.
Differential influences of shame and guilt on cooperation
Most people would agree that the experiences of guilt and shame are
unpleasant and not something that we would strive for. Still, experiences of
negative emotions can have positive consequences. For centuries economists
and psychologists have argued that moral emotions such as guilt and shame
lead to prosocial or cooperative behaviours (Frank, 1988; Ketelaar, 2004;
Smith, 1759). The idea is that when people feel guilty, they will try to make
up for the harm that they caused. Put differently, moral emotions motivate
people to act prosocially, which has positive consequences for the people
around them. However, the question is whether these positive effects are
present for all moral emotions. In this article we argue and show that
Correspondence should be addressed to: Ilona E. de Hooge, Department of Economic and
Social Psychology, Tilburg University, PO Box 90153, NL-5000 LE Tilburg, The Netherlands.
E-mail: I. E.deHooge@UvT.nl
We thank Paul van Lange for providing us with a programmed version of the ten-coin give-
some dilemma game, Eric van Dijk for comments on an earlier version of this manuscript, and
Linda de Hooge for help with the data collection of Experiment 2.
COGNITION AND EMOTION
2007, 21 (5), 1025 1042
#
2007 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business
www.psypress.com/cogemotion DOI: 10.1080/02699930600980874
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
prosocial effects in the short term are found for the moral emotion guilt but
not for shame, another moral emotion.
Moral emotions and cooperative behaviour
Moral emotions can be understood as emotions that are linked to the
interests or welfare of society as a whole or of other people (Haidt, 2003).
These emotions originate in social relationships and are built on reciprocal
evaluations and judgements of the self and others (Tangney & Fischer,
1995). Adam Smith, the founder of modern economics, stated as early as
1759 that moral emotions motivate cooperation. These emotions lead people
to focus on the other and on how one’s own behaviour affects the other’s
well-being. In situations where there is a conflict between the self-interest of
a person and interest of a group (a social dilemma), moral emotions are
claimed to motivate people to act in favour of other people’s interests
(Frank, 1988; Ketelaar, 2004; Smith, 1759). In this way, moral sentiments
motivate cooperative behaviours.
Frank (1988) has elaborated on the cooperative effect of moral emotions
in his commitment theory. People are often confronted with situations where
they have to choose between defection, which rewards the individual in the
short run but is costly for the group and the individual in the long run, and
cooperation, which is costly for the individual in the short run but beneficial
for the group and the individual’s self-interest in the long run. People’s
selfishness may seduce them to choose the attractive immediate reward at the
expense of long-term benefits. According to Frank (1988), emotions can act
as commitment devices that help us to resolve these social dilemmas.
Choosing for immediate individual rewards in a social dilemma situation
gives rise to negative moral emotions like guilt. These emotions make the
option of immediate individual rewards less attractive than the more
effective long-term strategy. In this way moral emotions commit people to
choose for the long-term strategy or for the group’s interest and thus
motivate cooperative behaviours.
Only recently Ketelaar and Au (2003) found empirical results that are
consistent with these claims. They studied the effects of guilt on cooperation.
Ketelaar and Au hypothesised that guilt would increase cooperation
especially for people with the general tendency to act uncooperatively.
These people (hereafter referred to as proselfs) would perceive their feelings
of guilt as a consequence of their negative behaviour and use this as
information about future costs of pursuing an uncooperative strategy. This
would lead them to act more cooperatively compared to proselfs who do not
experience guilt. People with the general tendency to act cooperatively
(hereafter referred to as prosocials) would already act cooperatively and thus
not use the negative feeling state as an inference about their strategy.
1026
DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
In their first study, Ketelaar and Au (2003) started with measuring the
general tendencies to act (un)cooperatively by letting participants play 40
rounds of a repeated social dilemma game (on the basis of their choices they
were classified as proselfs or prosocials). After these 40 rounds, an
autobiographical recall procedure followed, by which feelings of guilt were
induced. Participants were asked to give a detailed description of a recent
experience in which they felt very guilty. After writing for 10 minutes,
participants again played the same social dilemma game for 40 rounds. Their
choices in the first 10 rounds of these 40 rounds formed the dependent
measure of cooperative behaviour. The results showed that for proselfs
feelings of guilt led to more cooperative behaviour. The induction of guilt
had no effect on prosocials. In a second study, Ketelaar and Au found
similar results for naturally occurring guilt in a two-round ultimatum
bargaining game. Those who felt guilty over an unfair offer in the first round
were more likely to make a prosocial offer in the second round than those
who did not feel guilty.
These findings were recently replicated by Nelissen, Dijker, and De Vries
(2007), who studied the influences of fear and guilt on cooperation in a one-
shot give-some dilemma game. They hypothesised guilt mainly to motivate
cooperation for proselfs, as prosocials would already have cooperation
chronically accessible goal activated in cooperation situations. As hypothe-
sised, induction of guilt by an autobiographical recall procedure motivated
cooperation only for proselfs and not for prosocials. Contrary to guilt, the
induction of fear decreased cooperation for prosocials and not for proselfs.
Taken together, the findings of Ketelaar and Au (2003) and Nelissen et al.
(2007) show that the moral emotion guilt motivates cooperation. The
question that we address in this article is whether this effect on cooperative
behaviour generalises to other another moral emotion: shame. We have
reason to expect that shame does not produce cooperative tendencies.
Shame and guilt as two moral emotions
The emotion literature assumes that differential emotions have differential
influences on judgement and behaviour (e.g., Izard, 1993; Lerner & Keltner,
2000; Zeelenberg & Pieters, 2006). Emotions can be differentiated in terms of
feelings, thoughts, appraisals, action tendencies, and actions (e.g., Frijda,
1986; Roseman, Wiest, & Swartz, 1994). Specific emotions give information
about specific problems to be dealt with and motivate people to behave in
ways to solve the problem (Zeelenberg & Pieters, 2006). In this study we
focus on these motivational influences of specific emotions on behaviour.
Guilt is an emotion that arises after a moral transgression (Baumeister,
Stillwell, & Heatherton, 1994). People experiencing guilt have hurt someone
with their behaviour and perceive themselves as a bad person (Lewis, 1971).
MORAL SENTIMENTS 1027
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
What follows is the tendency to make up for the wrongdoing and to
undertake actions to minimise the damage caused (Tangney, Miller, Flicker,
& Barlow, 1996). Thus, the motivations that are associated with guilt are
consistent with the prosocial behaviours observed in the study of Ketelaar
and Au (2003). As cooperative behaviours are already chronically activated
in prosocials, the motivational effect of guilt on cooperation will primarily
be present in the behaviour of proselfs (Nelissen et al., 2007).
Shame is another moral emotion that is closely related to guilt. This
emotion arises after a moral transgression or after exposure of incompetence
(Keltner & Buswell, 1996). The person has shown to be inadequate and feels
worthless and inferior compared to others (Ausubel, 1955; Tangney, Wagner,
& Gramzow, 1992). When experiencing shame, the focus is on the self and
the general tendency of an ashamed person in the short term is to hide or
withdraw from the situation (Tangney et al., 1996). The differences between
guilt and shame can be found in Table 1.
The question we pose here is whether shame also motivates prosocial
behaviour. We pose this question because shame is associated with a focus on
the self and a motivation to hide or withdraw from the social situation.
These motivational tendencies are very different from those of guilt (i.e.,
reparation) and we argue that they are not logically related to cooperative
behaviour. Therefore, we expect that shame will not promote short-term
cooperation. Such a finding could have important consequences for theories
about moral emotions and cooperation. Accordingly, this finding would
contradict the general assumption that moral emotions unequivocally
motivate prosocial behaviours.
We conducted two experiments to test our hypotheses about the effects of
guilt and shame. Social motives were measured with the often used Triple
Dominance Measure of Social Value Orientations (Van Lange, Otten, De
TABLE 1
Differences between guilt and shame according to emotion literature
Emotion
Guilt Shame Reference
Eliciting event Moral transgression Moral transgression
or incompetence
Baumeister et al., 1994;
Keltner & Buswell, 1996
Appraisal Done damage Centre of attention Lewis, 1971; Tangney, 1991
Self-experience Bad person Weak person Lewis, 1971; Tangney &
Fischer, 1995
Action tendency Make up for
wrongdoing
Hide/withdraw Lindsay-Hartz, 1984;
Tangney et al., 1996
1028 DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
Bruin, & Joireman, 1997). In both experiments we first induced feelings of
shame or of guilt using an autobiographical recall procedure, similar to that
of Ketelaar and Au (2003). After this manipulation, participants in
Experiment 1 played a dyadic social dilemma game. They decided to what
extent they would act cooperatively when interacting with another person.
In Experiment 2, we assessed cooperation by means of a newly developed
everyday cooperation scale. In both experiments we expected and found guilt
to increase cooperative behaviour, especially for proselfs, as was found by
Ketelaar and Au (2003). For shame, we expected and found no effect on
cooperation.
EXPERIMENT 1
Method
Participants. Undergraduate economics and psychology students at
Tilburg University (66 males and 76 females, M
age
/20, SD /1.89) partici-
pated in this study in partial fulfilment of a course requirement. The study had
a 3 (Emotion condition: Control vs. Guilt vs. Shame)
/2 (SVO: Prosocial vs.
Proself) between-subjects factorial design with cooperation in a one-shot
social dilemma game as dependent variable. Participants were randomly
assigned to one of the three emotion conditions. Originally, 142 students
participated in this study, but 16 participants who could not be classified as
prosocial or proself were therefore left out of the analyses. In studies using this
measure of social motives it is common that 10 20% of the participants are
unclassifiable (Nelissen et al., 2007; Van Lange & Visser, 1999). Of the
participants, 71% of the males and 52% of the females were proselfs.
Procedure and variables. Participants entered the laboratory in groups of
eight to twelve participants. They were seated in separate cubicles and
informed that the hour consisted of multiple, unrelated studies. All the tasks
were unrelated and we were interested in whether emotion induction in one
task would influence cooperation in a subsequent unrelated task. The studies
consisted of computer tasks and paper-and-pencil tasks and were all in
Dutch. First, participants were asked to complete a questionnaire that was
placed next to the computer. This questionnaire was our emotion-induction
manipulation and was adopted from Ketelaar and Au (2003). We used the
Dutch emotion words ‘‘schuld’’ and ‘‘schaamte’’ for the English emotion
words ‘‘guilt’’ and ‘‘shame’’. Previous cross-cultural research showed that
these words in both languages refer to similar emotion experiences, emotion
elements and feelings (Breugelmans & Poortinga, 2006; Fontaine et al.,
2006). In the Guilt condition, participants were asked to report a personal
experience in which they felt very guilty. Participants wrote for example
MORAL SENTIMENTS 1029
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
about cheating on their romantic partner, forgetting a friend’s birthday,
breaking valuable things belonging to others and other related behaviours.
In the Shame condition, participants were asked to report a personal
experience in which they felt very ashamed. For example, they wrote about
bad performance in sports, giving a bad presentation, or failing an exam. In
the Control condition, participants were asked to describe a regular
weekday. Participants worked approximately 10 minutes on this emotion-
induction task.
Next, participants continued with an unrelated cooperation game, which
was our dependent measure. Participants played, via the computer, a ten-
coin give-some dilemma game with another participant (Van Lange &
Kuhlman, 1994). At the beginning of the game the participants had ten
coins, each worth t0.50 for the participant but t1 for the interaction partner.
The interaction partners also had ten coins, each worth t 0.50 for themselves
but t1 for the participant. The participant decided how many coins to give to
the interaction partner, without knowing how many coins the interaction
partner would give. In this game, participants would earn most together
when both offered all coins to the interaction partner (the cooperative
option). In contrast, the participants themselves would earn most when
keeping all their coins to themselves (the selfish option). The number of
coins offered was the measure of cooperation.
After the game ended participants continued with a third task. Here they
were asked to go back to the first questionnaire and to reread the description
they provided in the first task (the emotion-induction task). To check if the
emotion manipulation worked properly, participants subsequently indicated
how much shame and guilt they felt in the described situation. Because one
could experience a number of different emotions in these situations, we also
asked participants to indicate how much regret, disappointment, sadness,
fear, anger at self, anger at others, and dissatisfaction they felt in the
situation. All emotions were rated on an 11-point scales ranging from 0 (not
at all)to10(very strongly).
The task including the manipulation checks was followed by multiple,
unrelated computer tasks and paper-and-pencil studies. At the end of the
hour, participants’ social value orientation (SVO) was assessed via the Triple
Dominance Measure (Van Lange et al., 1997). Usually, two types of SVO are
distinguished: prosocials (who maximise joint gain and strive for equality)
and proselfs (who maximise their own outcome). Participants were classified
as prosocials (n
/49) or proselfs (n/77) based on at least six (out of nine)
consistent choices. Importantly, the proself/prosocial classification was
unaffected by the emotion induction, x
2
(2, n/126)/0.56, ns. After
completion of all tasks participants were thanked and debriefed.
1030
DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
Results
Manipulation checks. Results of both the manipulation checks and
cooperation are displayed in Table 2. The manipulation checks showed that
the emotion induction was successful. Participants in the Guilt condition
reported significantly more guilt than participants in the Shame condition,
t(85)
/5.37, pB/.001, and participants in the Control condition, t(79)/
14.27, pB/.001. Participants in the Shame condition reported significantly
more shame than participants in the Guilt condition, t(85)
/3.77, pB/.001,
and participants in the Control condition, t(82)
/26.94, p B/.001. Further-
more, participants in the Guilt condition felt more guilty than ashamed,
t(41)
/4.97, p B/.001, and participants in the Shame condition felt more
ashamed than guilty, t(44)
/6.47, p B/.001. There were no differences
between the Guilt and Shame conditions on the other emotions assessed,
all ts(85) B
/1.76. The emotion manipulation worked for both proselfs and
prosocials, as all reported tests were also significant for prosocials and
proselfs.
Cooperation. We tested two hypotheses. First, we hypothesised guilt to
increase cooperation especially for proselfs, replicating the findings of
Ketelaar and Au (2003). We expected participants in the Guilt condition
(especially proselfs) to contribute more to the other person than participants
TABLE 2
Means (standard deviations) of experienced emotion and cooperation in
Experiment 1
Emotion condition
Control (n
/39) Guilt (n /42) Shame (n/45)
Experienced emotion M (SD) M (SD) M (SD)
Guilt 1.87 (1.89) 8.60 (1.53) 5.38 (3.59)
Prosocial 1.91 (2.35) 8.65 (1.39) 5.12 (3.84)
Proself 1.81 (3.02) 8.55 (1.68) 5.74 (3.28)
Shame 1.10 (1.43) 7.57 (1.89) 8.84 (1.21)
Prosocial 1.30 (1.49) 7.60 (1.70) 9.00 (1.20)
Proself 0.81 (1.33) 7.55 (2.09) 8.63 (1.21)
Cooperation
Prosocials 5.24 (3.70) 4.93 (3.20) 5.53 (2.27)
Proselfs 3.41 (3.94) 6.04 (3.04) 3.04 (3.31)
Note : Guilt and Shame were measured on an 11-point scale, ranging from 0 (not at all )to10(very
strongly ). Cooperation reflects the number of coins (out of 10) that the participant donated to the
other person in the dyad. A higher score signifies higher cooperation.
MORAL SENTIMENTS 1031
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
in the Control condition. Second, we hypothesised shame to have no effect
on cooperation. We expected participants in the Shame condition to
contribute the same amount of coins as participants in the Control
condition.
The findings supported our hypotheses. A 3 (Emotion condition)
/2
(SVO) ANOVA on the number of coins contributed to the other player
showed only a significant Emotion
/SVO interaction, F(2, 120) /3.33, p B/
.05, h
p
2
/.05. The effects of emotion condition on cooperation differed for
prosocials and proselfs. For proselfs, guilt feelings had a significant influence
on cooperation. Proselfs in the Guilt condition contributed significantly
more than proselfs in the Control condition, t(47)
/2.64, p B/.05, and
proselfs in the Shame condition, t(53)
/3.50, p B/.001. Proselfs in the Shame
condition did not contribute significantly more than proselfs in the Control
condition, t(48)
/0.37, ns. If anything, proselfs in the Shame condition
contributed less than participants in the Control condition. For prosocials,
there was no significant difference between the Guilt condition and the
Control condition, t(30)
/0.25, ns, and between the Guilt condition and the
Shame condition, t(30)
/0.61, ns. Also, contributions of prosocials in the
Shame condition did not differ significantly from contributions of prosocials
in the Control condition, t(32)
/0.28, ns. These results could not be
explained by Gender, as a 3 (Emotion condition)
/2 (Gender) ANOVA
showed no significant results, F(2, 120)
/0.43, ns. Thus, guilt motivated
cooperation only for proselfs whereas shame did not have any influence on
cooperation.
Discussion
Contrary to the assumption of scholars like Smith (1759) and Frank (1988),
we showed in this study that not all moral emotions motivate cooperation.
Using a procedure similar to that of Ketelaar and Au (2003) and Nelissen
et al. (2007), we found that guilt increased cooperation in a dyadic social
dilemma game especially for proselfs. Shame, another moral emotion, did
not motivate cooperation in a dyadic social dilemma.
It is important to note that the effect of guilt on cooperation is a perfect
replication of Ketelaar and Au (2003), even though we used different
measures. While they used the first 40 rounds of a repeated bargaining game
to assess the social motives of their participants, we used an independent and
validated measure to classify participants as prosocials or proselfs (Van
Lange et al., 1997). In addition, Ketelaar and Au used the first 10 rounds of
the second set of 40 rounds of the repeated social dilemma game to measure
cooperation, while we used a single-shot give-some dilemma game that was
clearly unrelated to the SVO measure. Despite these procedural differences,
the cooperative effect of guilt on behaviour for proselfs was replicated. We
1032
DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
consider this a very valuable replication, because until now the results of
Ketelaar and Au (2003) and Nelissen et al. (2007) constituted the sole
support for increased cooperation after guilt. The effect of guilt on
cooperation appears to be fairly robust.
In contrast to our findings for guilt, we found no evidence for increased
cooperation in the social dilemma game after induction of shame. If
anything, this emotion led to a (nonsignificant) decrease in cooperation in
the social dilemma. At least in the situations examined here, this finding
suggests that not all moral emotions motivate cooperative behaviour.
There are three alternative explanations for our results that we cannot
rule out on the basis of the data obtained in Experiment 1. First, it could be
that the induction of shame was less successful or less intense than the
induction of guilt. The emotion manipulation check in Experiment 1 does
not preclude this, because we asked participants to indicate the intensity of
the emotions experienced in the described situation and not the emotions
experienced at that moment. Thus, we do not know whether autobiogra-
phical recall had similar effects for shame as for guilt on current emotional
experience.
A second reason why our induction of shame might not have been
successful is the time elapsed between the recalled event and the moment in
which the experiment took place. Rime´, Mesquita, Philippot, and Boca
(1991), for example, found that shame events took place longer ago than
guilt events in their study on the social sharing of emotions. This may imply
that the shame events reported in our study could have taken place longer
ago than the guilt events. If this is the case, shame experiences could have
been less intense than guilt experiences and therefore would have no effect on
cooperation. We do not think that the results of Rime´ et al. can be easily
generalised to our study, because they investigated emotional events that
people tended to talk about with others and these may differ from the one
that people write about in the task we used. But, because we did not ask for
the time elapsed since the event, we cannot rule out this possibility.
A third reason for the non-effects of shame may be that our measure
of cooperation was not sensitive to the effects of shame. We measured
cooperation in a one-shot, dyadic social dilemma game. It is possible that the
effects of shame on cooperation can not be picked up by social dilemma
games. Other measures for cooperation could show whether our results in
Experiment 1 can be generalised to other cooperation settings.
In order to address these points, we conducted a second experiment. In
Experiment 2, shame and guilt were again induced using the autobiogra-
phical recall procedure. Following the emotion induction, participants
reported the emotions they were feeling at that very moment. To control
for possible differences in shame and guilt in time elapsed between the
moment the reported event took place and the experiment, we asked
MORAL SENTIMENTS 1033
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
participants to indicate the time elapsed since the event took place. Finally,
to see whether the results of Experiment 1 can be generalised outside social
dilemma games, we measured cooperation with the Everyday Cooperation
Scale, a 9-item scale considering cooperation in daily situations. We expected
a replication of Experiment 1 for the effects of both guilt and shame.
EXPERIMENT 2
Method
Participants. Undergraduate students at Avans University Breda and at
Tilburg University (78 males and 73 females, M
age
/21, SD/2.72)
volunteered to participate in this study. They were randomly assigned to
one of the three emotion conditions of the 3 (Emotion condition: Control vs.
Guilt vs. Shame)
/2 (SVO: Prosocial vs. Proself) between-subjects factorial
design. Originally, one hundred and fifty one students participated in this
study, but thirty three participants who could not be classified as prosocial
or proself were left out of analyses. Of the participants, 52% of the males and
35% of the females were proselfs.
Procedure and variables. Participants were seated and informed that the
questionnaire they would fill in consisted of multiple, unrelated studies.
Again, all the tasks were unrelated and we were interested in whether
emotion induction in task one would influence cooperation in task two.
First, depending upon the condition, participants were asked to report a
personal experience in which they felt very guilty (Guilt condition) or very
ashamed (Shame condition), or to describe a regular weekday (Control
condition). This was the same manipulation as in Experiment 1.
After describing the personal experience, participants were asked to
indicate the time elapsed since this event took place. They indicated how
many weeks, months and years ago the event took place. The reported time
elapsed was converted to weeks. Then, to check whether the emotion
manipulation worked properly, participants rated how much shame and guilt
they felt at that very moment. We also asked participants to rate how much
regret, disappointment, sadness, fear, anger at self, anger at others, and
dissatisfaction they felt at that moment. All emotions were rated on an 11-
point scales ranging from 0 (not at all)to10(very strongly).
Next, participants continued with a new task, a cooperation question-
naire that was our dependent measure. This questionnaire was a specially
developed measure of general cooperation tendencies, hereafter referred to
as the Everyday Cooperation Scale. This 9-item state measure was inspired
by a 23-item trait cooperation measure, the Prosocial Tendencies Measure
(Carlo & Randall, 2002). From the Prosocial Tendencies Measure we left out
1034
DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
the items concerning compliant helping (that is, helping others in response to
a request), items concerning money or goods, and items concerning helping
in order to enhance one’s own position because these items did not reflect
everyday, voluntary cooperation. A factor analysis on the nine items showed
a clear one-factor solution (see the Appendix for the items and factor
loadings). This factor had an eigenvalue of 4.85, explained 50% of the
variance, and the nine items formed a reliable scale (a
/.89). For each item,
participants were asked to report how much they wanted to undertake that
action at that very moment. All items were rated on an 11-point scales
ranging from 0 (not at all)to10(very much).
At the end of the questionnaire, participants’ social value orientation
(SVO) was assessed via the Triple Dominance Measure (Van Lange et al.,
1997). Participants were classified as prosocials (n
/66) or proselfs (n/52)
based on at least six (out of nine) consistent choices. This classification was
unaffected by the emotion induction, x
2
(2, n/118)/0.39, ns. After
completion of all tasks participants were thanked and debriefed.
Results
Manipulation checks. Results of both the manipulation checks and
cooperation are displayed in Table 3. The manipulation checks showed that
the emotion inductions were successful. Participants in the Guilt condition
TABLE 3
Means (standard deviations) of experienced emotion and cooperation in
Experiment 2
Emotion condition
Control (n
/37) Guilt (n /44) Shame (n/48)
Experienced emotion M (SD) M (SD) M (SD)
Guilt 0.41 (0.69) 6.11 (2.10) 5.08 (2.73)
Prosocial 0.41 (0.67) 6.42 (2.02) 5.06 (2.69)
Proself 0.40 (0.74) 5.84 (2.22) 5.13 (2.90)
Shame 0.62 (1.04) 5.25 (2.50) 6.65 (2.09)
Prosocial 0.73 (1.16) 5.42 (2.67) 6.75 (1.97)
Proself 0.47 (0.83) 4.95 (2.35) 6.44 (2.34)
Cooperation
Prosocials 6.72 (1.21) 6.06 (0.90) 6.26 (1.12)
Proselfs 5.31 (1.90) 6.36 (0.78) 5.41 (1.44)
Note : Guilt and Shame were measured on an 11-point scale, ranging from 0 (not at all )to10(very
strongly ). Cooperation reflects the mean score on the nine cooperation items, ranging from 0 (not at
all )to10(very much ).
MORAL SENTIMENTS 1035
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
reported significantly more guilt than participants in the Shame condition,
t(90)
/2.02, p B/.05, and participants in the Control condition, t(79)/15.80
pB
/.001. Participants in the Shame condition reported significantly more
shame than participants in the Guilt condition, t(90)
/2.92, pB/.01, and
participants in the Control condition, t(83)
/10.18, pB/.001. More im-
portantly, participants in the Guilt condition felt more guilty than ashamed,
t(43)
/2.98, p B/.01, and participants in the Shame condition felt more
ashamed than guilty, t(47)
/4.18, p B/.001. There were no differences
between Guilt and Shame conditions on the other assessed emotions, all
ts(90)B
/1.61. The emotion manipulation worked for both proselfs and
prosocials, as all reported tests were also significant for prosocials and
proselfs.
There was no difference between reported time elapsed in the Guilt
condition (M
/117, SD/198) and the Shame condition (M/91, SD /
115), t(90)/0.70, ns. Furthermore, the reported time elapsed had no
influence on the reported guilt, b
/.12, ns, no influence on reported shame,
b
/.14, ns, and no influence on cooperation, b /.11, ns.
Cooperation. This experiment again tested two hypotheses. First, we
hypothesised guilt to increase cooperation especially for proselfs, replicating
Experiment 1 and the findings of Ketelaar and Au (2003). We expected
proselfs in the Guilt condition to have a higher score on the cooperation
scale than proselfs in the Control condition. Second, also replicating
Experiment 1, we hypothesised shame to have no effect on cooperation.
We expected participants in the Shame condition to have the same score on
the cooperation scale as participants in the Control condition.
The findings supported our hypotheses. As predicted, a 3 (Emotion
condition)
/2 (SVO) ANOVA showed only a significant Emotion/SVO
interaction, F(2, 99)
/4.02, pB/.05, h
p
2
/.08. The effects of emotion
condition on cooperation differed for prosocials and proselfs. For proselfs,
guilt feelings had a significant influence on cooperation. Proselfs in the Guilt
condition had a significantly higher score than proselfs in the Control
condition, t(29)
/1.97, p/.05, and proselfs in the Shame condition, t(28)/
2.24, pB/.05. Proselfs in the Shame condition did not have a significantly
different score compared to proselfs in the Control condition, t(29)
/0.16,
ns. For prosocials, there was no significant difference between the Guilt
condition and the Control condition, t(36)
/1.92, ns, and between the Guilt
condition and the Shame condition, t(36)
/0.64, ns. Also, scores of
prosocials in the Shame condition did not differ significantly from scores
of prosocials in the Control condition, t(40)
/1.30, ns. These results could
not be explained by Gender, as a 3 (Emotion condition)
/2 (Gender)
ANOVA showed no significant results, F(2, 99)
/2.04, ns. Thus, again, guilt
1036
DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
motivated cooperation for proselfs whereas shame did not have any
influence on cooperation.
Discussion
In this experiment, we fully replicated the findings of Experiment 1. Guilt
again motivated cooperation for proselfs, this time on the Everyday
Cooperation Scale. In contrast, shame, the other moral emotion, did not
have any influence on cooperation in everyday situations.
These results showing the positive effect of guilt on cooperation and the
non-effect of shame on cooperation cannot be explained by a less successful
or less intense induction of shame compared to guilt, neither by the time
elapsed since the reported event took place. Both of these were possible
alternative explanations of the absence of an effect of shame in Experiment
1.We consider the most important finding the replication for both guilt and
shame on a totally different cooperation measure, our specially developed
Everyday Cooperation Scale. This makes our findings from Experiment 1
more reliable and shows that the effects of guilt and shame on cooperation
can be generalised to different situations.
GENERAL DISCUSSION
Economists and psychologists like Frank (1988), Ketelaar (2004), and Smith
(1759) assumed that moral emotions motivate cooperation. In two studies
we have shown that this claim is not applicable to all moral emotions. While
the moral emotion guilt motivated cooperation in both social dilemma
situations and everyday situations, shame, another moral emotion, did not
have an effect on cooperation in either of these measures. Thus, not all moral
emotions motivate cooperative behaviour.
The differential effects of guilt and shame on cooperation can be
explained by inspecting the motivations that accompany the emotions
(Zeelenberg, Nelissen, & Pieters, 2006). We think that specific emotions
motivate people to behave in different ways, leading to different behaviours.
Guilt signals that one has hurt another person and motivates reparative
behaviour in order to undo the wrongdoing (Tangney et al., 1996). As
prosocial or cooperative behaviour is a way to repair, this emotion motivates
cooperation in social dilemmas and in everyday situations. Shame signals
that one has made a mistake and motivates withdrawal in the short term in
order to avoid more mistakes (Tangney et al., 1996). As cooperation or
prosocial behaviour is not a withdrawal strategy, shame will not have any
influence on cooperation in the short term.
The effect of guilt on cooperation appears to be fairly robust. Ketelaar
and Au (2003), Nelissen et al. (2007) and our Experiment 1 found guilt to
MORAL SENTIMENTS 1037
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
motivate cooperation in social dilemma games. We replicated this finding
with an everyday cooperation measure in Experiment 2. This effect of guilt
can play an important role in economic behaviour. While economic theory
assumes that people act according to immediate self-interest, guilt is a moral
emotion that can easily motivate people to act more cooperatively.
The findings can also play an important role in emotion research. The
present studies show that in addition to the variables measured in traditional
emotion research, guilt and shame can be differentiated on the basis of the
behavioural consequences that ensue from the emotions. Here, emotion
researchers can benefit by adopting a decision-making perspective. Decision
researchers have traditionally focused on the effects of variables on
behavioural choice, and our research shows how some of their paradigms
can be useful in differentiating emotions.
Like Ketelaar and Au (2003) and Nelissen et al. (2007), we only found a
cooperative effect of guilt for proselfs. One may argue that in prosocials the
motivation to act cooperatively is chronically activated (Nelissen et al.,
2007), hence no additional activation of this motivation by guilt is to be
expected. In contrast, for proselfs a cooperative motivation is not chronically
activated. For these people, the experience of guilt activates this motivation
and subsequently increases cooperation. As shame does not activate a
motivation to cooperate, there are no differential effects of this emotion for
proselfs and prosocials.
Before closing, we would like to make three observations regarding the
specifics of our experiments that call for future research. The first
observation pertains to the reported emotions in the Shame and Guilt
conditions. The amount of guilt reported in the Shame condition is lower
than the amount of reported shame, but still greater than the amount of guilt
reported in the Control condition. If guilt has a separate bearing on
cooperative behaviour then we would have expected proselfs to have shown
at least some increase in cooperation in the Shame condition in comparison
to the Control condition. The fact that we did not find such an effect may
suggest that, even though we may experience multiple emotions at the same
time, our behaviours are motivated by a single emotion at that time. There
may be several explanations for this finding. It is possible that the strongest
emotion cancels out the action tendencies of any other emotions, leaving
only shame to influence cooperation in the Shame condition. It is also
possible that the strongest emotion gets action priority (see Frijda, 1986)
whereas less intense emotions, when not faded yet, can influence behaviour
after the strongest emotion has been reacted upon. This would mean that
guilt could still affect cooperative behaviours after the effects of shame have
faded. There is a clear need for future research considering how multiple
experienced emotions influence our behaviour.
1038
DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
A second observation concerns the used measures for cooperation. The
replication of our results on these two very different measures suggests that
our findings can be considered fairly robust. However, the finding that
shame does not motivate cooperation on these measures does not necessarily
imply that shame does not motivate cooperation in general. We only
considered short-term cooperation and not long-term cooperation. It is
possible that shame motivates cooperation in the long term, for example by
conformity to group norms (see Fessler, 2004). Further empirical research is
needed to investigate whether the long-term effects of shame on cooperation
are similar to the short-term effects of shame.
A third observation relates to the emotion manipulation used in our
studies. In both studies, we induced emotions with an autobiographical
recall procedure. This procedure was also used by Ketelaar and Au (2003),
Nelissen et al. (2007) and others. In such a procedure, the emotional state
that is elicited is not directly related to the following task (in our case the
dyadic interaction and the cooperation scale). Zeelenberg and Pieters (2006)
term emotions in such situations as ‘‘exogenous’’, to emphasise the fact that
the emotional influence comes from outside the relevant situation. In our
studies, we assume that the motivation activated by the emotion influences
behaviour across situations. This is supported by our findings that
exogenous guilt motivated cooperation, corresponding to the motivation
to repair, and that exogenous shame did not influence cooperation,
corresponding to the motivation to withdraw. As yet it is not clear whether
these exogenous effects of emotions are always similar to their endogenous
effects (that is, when the emotion is directly relevant for the task at hand).
One study supporting the assumption that exogenous and endogenous
effects of emotions are similar is the second study of Ketelaar and Au (2003).
The results of this study showed that endogenous guilt had the same effects
on cooperation as exogenous guilt. This finding gives us support for the
assumption that our results also apply to endogenous guilt. There is,
however, no research considering the effects of endogenous shame. It may be
the case that experiences of shame that are directly relevant for a current
social interaction do have an effect on that interaction. For example, some
social dilemma games include an exit option. This option offers people the
possibility of avoiding the interaction situation altogether without the
would-be interaction partner ever knowing about the possibility of inter-
acting (Dana, Cain, & Dawes, 2006). We would expect shame to have effects
in such a situation, because of its motivational tendency to withdraw from
the situation. It may also be the case that exogenous shame does not
motivate cooperation, but that endogenous shame does motivate coopera-
tion. We are currently exploring the effects of endogenous shame on
cooperation.
MORAL SENTIMENTS 1039
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
In closing, let us return to the question that motivated our research,
namely whether moral emotions in general promote cooperative behaviour.
We have obtained clear support for the idea that different moral emotions
may have specific effects on cooperative behaviour. As such, this research
demonstrated the use of thinking about emotions in motivational terms and
the importance of emotion-specific predictions, even when considering
emotions that are as closely related as shame and guilt.
Manuscript received 22 December 2005
Revised manuscript received 15 August 2006
Manuscript accepted 24 August 2006
REFERENCES
Ausubel, D. P. (1955). Relationships between shame and guilt in the socialization process.
Psychological Review, 67, 378 390.
Baumeister, R. F., Stillwell, A. M., & Heatherton, T. F. (1994). Guilt: An interpersonal
approach. Psychological Bulletin, 115, 243 267.
Breugelmans, S. M., & Poortinga, Y. H. (2006). Emotion without a word: Shame and guilt with
Rara´muri Indians and rural Javanese. Journal of Personality and Social Psychology, 91,
1111 1122.
Carlo, G., & Randall, B. A. (2002). The development of a measure of prosocial behaviors for late
adolescents. Journal of Youth and Adolescence, 31,3144.
Dana, J., Cain, D., & Dawes, R. (2006). What you don’t know won’t hurt me: Costly (but quiet)
exit in dictator games. Organizational Behavior and Human Decision Processes, 100, 193
201.
Fessler, D. M. T. (2004). Shame in two cultures: Implications for evolutionary approaches.
Journal of Cognition and Culture, 4, 207 262.
Fontaine, J. R. J., Luyten, P., De Boeck, P., Corveleyn, J., Fernandez, M., Herrera, D., et al.
(2006). Untying the Gordian knot of guilt and shame: The structure of guilt and shame
reactions based on situation and person variation in Belgium, Hungary, and Peru. Journal of
Cross-Cultural Psychology, 37, 273 292.
Frank, R. H. (1988). Passions within reason: The strategic role of the emotions. New York:
Norton.
Frijda, N. H. (1986). The emotions. Cambridge, UK: Cambridge University Press.
Haidt, J. (2003). The moral emotions. In R. J. Davidson, K. R. Scherer, & H. H. Goldsmith
(Eds.), Handbook of affective sciences (pp. 852 870). Oxford, UK: Oxford University Press.
Izard, C. E. (1993). Organizational and motivational functions of discrete emotions. In M.
Lewis & J. M. Haviland (Eds.), Handbook of emotions (pp. 631 642). New York: Guilford
Press.
Keltner, D., & Buswell, B. N. (1996). Evidence for the distinctness of embarrassment, shame and
guilt: A study of recalled antecedents and facial expressions of emotion. Cognition and
Emotion, 10, 155 171.
Ketelaar, T. (2004). Ancestral emotions, current decisions: Using evolutionary game theory to
explore the role of emotions in decision-making. In C. Crawford & C. Salmon (Eds.),
Evolutionary psychology, public policy and personal decisions (pp. 145 163). Mahwah, NJ:
Lawrence Erlbaum Associates, Inc.
1040 DE HOOGE, ZEELENBERG, BREUGELMANS
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
Ketelaar, T., & Au, W. T. (2003). The effects of guilt on the behaviour of uncooperative
individuals in repeated social bargaining games: An affect-as-information interpretation of
the role of emotion in social interaction. Cognition and Emotion, 17, 429 453.
Lerner, J. S., & Keltner, D. (2000). Beyond valence: Toward a model of emotion-specific
influences on judgment and choice. Cognition and Emotion, 14, 473 493.
Lewis, H. B. (1971). Shame and guilt in neurosis. New York: International Universities Press.
Lindsay-Hartz, J. (1984). Contrasting experiences of shame and guilt. American Behavioral
Scientist, 27, 689 704.
Nelissen, R. M. A., Dijker, A. J. M., & De Vries, N. K. (2007). How to turn a hawk into a dove
and vice versa: Interactions between emotions and goals in a give-some dilemma game.
Journal of Experimental Social Psychology, 43, 280 286 .
Rime´, B., Mesquita, B., Philippot, P., & Boca, S. (1991). Beyond the emotional event: Six studies
on the social sharing of emotion. Cognition and Emotion, 5, 435 465.
Roseman, I. J., Wiest, C., & Swartz, T. S. (1994). Phenomenology, behaviors, and goals
differentiate discrete emotions. Journal of Personality and Social Psychology, 67, 206 221.
Smith, A. (1759). The theory of moral sentiments. London: A. Miller.
Tangney, J. P. (1991). Moral affect: The good, the bad, and the ugly. Journal of Personality and
Social Psychology, 61, 598 607.
Tangney, J. P., & Fischer, K. W. (1995). Self-conscious emotions: The psychology of shame, guilt,
embarrassment, and pride. New York: Guilford Press.
Tangney, J. P., Miller, R. S., Flicker, L., & Barlow, D. H. (1996). Are shame, guilt and
embarrassment distinct emotions? Journal of Personality and Social Psychology, 70, 1256
1269.
Tangney, J. P., Wagner, P., & Gramzow, R. (1992). Proneness to shame, proneness to guilt, and
psychopathology. Journal of Abnormal Psychology, 101, 469 478.
Van Lange, P. A. M., & Kuhlman, D. M. (1994). Social value orientations and impressions of
partner’s honesty and intelligence: A test of the might versus morality effect. European
Journal of Social Psychology, 67, 126 141.
Van Lange, P. A. M., Otten, W., De Bruin, E. N. M., & Joireman, J. A. (1997). Development of
prosocial, individualistic, and competitive orientations: Theory and preliminary evidence.
Journal of Personality and Social Psychology, 21, 273 292.
Van Lange, P. A. M., & Visser, K. (1999). Locomotion in social dilemmas: How people adapt to
cooperative, tit-for-tat, and noncooperative partners. Journal of Personality and Social
Psychology, 77, 762 773.
Zeelenberg, M., Nelissen, R., & Pieters, R. (2006). Emotion, motivation and decision making: A
feeling is for doing approach. In H. Plessner, C. Betsch, & T. Betsch (Eds.), A new look on
intuition in judgment and decision making. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
Zeelenberg, M., & Pieters, R. (2006). Feeling is for doing: A pragmatic approach to the study of
emotions in economic behavior. In D. De Cremer, M. Zeelenberg, & K. Murnighan (Eds.),
Social psychology and economics (pp. 117 137). Mahwah, NJ: Lawrence Erlbaum
Associates, Inc.
MORAL SENTIMENTS
1041
Downloaded By: [Universiteit Van Tilburg] At: 10:47 31 July 2007
APPENDIX
Items and factor loadings of the everyday cooperation scale used in Experiment 2
(
n
/117, a/.89)
Item Factor loading
1. Help an unknown other .72
2. Help a person while others are looking at me .70
3. Comfort someone who is emotionally very upset .68
4. Help a person when (s)he does not know who is helping .72
5. Help a person while I get in the spotlight as a consequence .59
6. Support a person who is emotionally distressed .74
7. Help a person without him/her knowing .59
8. Help a person while others are watching the way I do everything .68
9. Help someone who hurt him/herself .60
Note : Items were complements to the sentence ‘‘At this moment I would like to ... ’’ and could be
answered on 11-point scales with end points labelled 0 (not at all ) and 10 (very much ).
1042 DE HOOGE, ZEELENBERG, BREUGELMANS
