It pays to be nice, but not really nice: Asymmetric evaluations of prosociality across seven cultures.

Abstract

Cultures differ in many important ways, but one trait appears to be universally
valued: prosociality. For one’s reputation, around the world, it pays to be nice to others.
However, recent research with American participants finds that evaluations of prosocial
actions are asymmetric—relatively selfish actions are evaluated according to the
magnitude of selfishness but evaluations of relatively generous actions are less sensitive
to magnitude. Extremely generous actions are judged roughly as positively as modestly
generous actions, but extremely selfish actions are judged much more negatively than
modestly selfish actions (Klein & Epley, 2014). Here we test whether this asymmetry in
evaluations of prosociality is culture-specific. Across 7 countries, 1,240 participants
evaluated actors giving various amounts of money to a stranger. Along with relatively
minor cross-cultural differences in evaluations of generous actions, we find cross-cultural
similarities in the asymmetry in evaluations of prosociality. We discuss implications for
how reputational inferences can enable the cooperation necessary for successful societies.

Full text

Judgment and Decision Making, Vol. 10, No. 4, July 2015, pp. 355–364
It pays to be nice, but not really nice:
Asymmetric reputations from prosociality across 7 countries
Nadav Klein∗Igor Grossmann†Ayse K. Uskul‡Alexandra A. Kraus§Nicholas Epley∗
Abstract
Cultures differ in many important ways, but one trait appears to be universally valued: prosociality. For one’s reputation,
around the world, it pays to be nice to others. However, recent research with American participants finds that evaluations
of prosocial actions are asymmetric—relatively selfish actions are evaluated according to the magnitude of selfishness but
evaluations of relatively generous actions are less sensitive to magnitude. Extremely generous actions are judged roughly as
positively as modestly generous actions, but extremely selfish actions are judged much more negatively than modestly selfish
actions (Klein & Epley, 2014). Here we test whether this asymmetry in evaluations of prosociality is culture-specific. Across
7 countries, 1,240 participants evaluated actors giving various amounts of money to a stranger. Along with relatively minor
cross-cultural differences in evaluations of generous actions, we find cross-cultural similarities in the asymmetry in evaluations
of prosociality. We discuss implications for how reputational inferences can enable the cooperation necessary for successful
societies.
Keywords: prosociality, selfishness, generosity, social judgment, reputation, culture, supererogation.
1 Introduction
Societies reveal their values through the behaviors they
praise and punish. Although societies may vary markedly,
most appear to highly value one fundamental trait in others:
prosociality. Selfless actions are publicly praised around
the world, such as Warren Buffet’s contractual commitment
to donate 99 percent of his wealth to charity and by Ma-
hatma Gandhi and Mother Teresa’s lifetime of self-sacrifice
for others. In Christianity, generosity is exalted as a spiri-
tual virtue. In Buddhism, generosity is likewise considered
one of the two characteristics necessary for enlightenment.
Western and Eastern philosophies both consider generosity
to be a virtue and a goal for one’s moral development, as
the writings of both Aristotle and Confucius reveal. In lit-
erature, Charles Dickens’ A Christmas Carol (1843) is as
popular in the Western hemisphere as the Chinese children’s
We thank Anna Leontieva for helping with data collection in Rus-
sia, Dmitrij Agroskin for helping with data collection in Austria, and
Haotian Zhou for helping with data collection in China. We thank the
Booth School of Business and the Insight grant from the Social Science
and Humanities Research Council of Canada for financial support. We
thank Jon Baron and an anonymous reviewer for helpful comments on this
manuscript. Correspondence concerning this article should be addressed
to Nadav Klein or Nicholas Epley, 5807 South Woodlawn Avenue, Booth
School of Business, University of Chicago, Chicago, IL 60637. E-mail:
nklein@chicagobooth.edu or epley@chicagobooth.edu.
Copyright: © 2015. The authors license this article under the terms of
the Creative Commons Attribution 3.0 License.
∗University of Chicago
†University of Waterloo
‡University of Kent
§Aarhus University
story Kong Rong Giving Up Pears (1778/2011)—a story
about a boy sharing his pears with his older siblings—is in
the Eastern hemisphere. And in one of the largest cross-
cultural studies focusing on gender differences in mate pref-
erences ever conducted (Buss, 1989), researchers neverthe-
less found a striking similarity: the prosocial trait of “kind-
understanding” was consistently among the most highly val-
ued traits by both genders in all cultures. For one’s reputa-
tion in the mind of others, around the world, it pays to be
nice.
Recent research, however, suggests that it may not pay
markedly more for one’s reputation to be really nice. That
is, whereas increasingly selfish behavior is judged increas-
ingly negatively by others, increasingly selfless behavior—
actions that benefit others more than the self—is not judged
markedly more positively by others. Instead of a mono-
tonic increase in evaluations across the entire spectrum of
prosocial behaviors ranging from completely selfish to com-
pletely selfless, there appears to be an asymmetry in evalu-
ations of relatively selfish versus selfless behavior. In one
experiment (Klein & Epley, 2014, Experiment 1a), concert-
goers judged another person who donated less than the sug-
gested donation amount for the concert to be less warm (e.g.,
less sincere, good-natured, and caring) than someone who
donated the suggested amount, but did not judge a person
who gave more than the suggested amount any more favor-
ably than the person who gave only the suggested amount.
In another experiment (Klein & Epley, 2014, Experiment
4a), participants evaluated a person who kept money for
himself from a bag found on the street increasingly more
negatively as the person kept an increasingly larger share of
355

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
356
the money before turning it into the police. Participants did
not, however, judge a person who gave away award money
to a charity increasingly more positively as the person gave
away an increasingly larger share of the money. Increasing
selfishness led to an increasingly negative reputation, but in-
creasing selflessness did not lead to an increasingly positive
reputation. In another experiment (Gray, Ward & Norton,
2014), people indirectly reciprocated either a selfish or eq-
uitable action from another person in kind, but did not recip-
rocate another person’s generous action with an equivalent
degree of generosity. Instead, they reciprocated a generous
action with a merely fair action. Selfishness was repaid in
kind measure, but selflessness was not. Additional research
similarly finds that actions that go beyond equitable distribu-
tions are often not evaluated any more positively than equi-
table actions (Loewenstein, Thompson & Bazerman, 1989;
Vesleý, 2015). These results suggest an important asymme-
try in the reputational value of prosocial behavior, such that
increasing selfishness leads to an increasingly negative eval-
uations but increasing selflessness does not lead to an equiv-
alent increase in positive evaluations. It pays to be nice, but
not really nice.
These results reflect more than ceiling effects (whereby
people want to evaluate selflessness more positively than
fairness but are artificially limited by a bounded measure-
ment scale) because similar results are obtained in un-
bounded measures, such as estimations of a person’s annual
charitable donations. Rather, these results reflect relative in-
sensitivity to magnitude when evaluating generous actions
but high sensitivity to magnitude when evaluating selfish
actions. When prosocial actions were judged in compari-
son to each other rather than in isolation, the asymmetry
in evaluations disappeared and increasingly generous actors
were evaluated increasingly favorably (Klein & Epley, 2014,
Experiments 4b and 5). These results suggest that people
can appreciate increasing generosity in others when differ-
ent levels of generosity are explicitly compared against each
other, but that judgments of prosocial actions in isolation
do not elicit these spontaneous comparisons and therefore
do not reflect this appreciation. These results seem to re-
flect a basic pattern of human judgment in which evaluations
are sensitive to magnitude, or scope, when they elicit com-
parisons to similar alternatives but are insensitive to scope
when they do not (Hsee & Zhang, 2010). A selfish action,
according to evidence from Klein and Epley (2014), enables
a person to think of more or less selfish counterfactuals and
thereby keep a given selfish action in perspective. A selfless
action, in contrast, does not seem to elicit the same kind of
comparative thinking, rendering evaluations of selfless ac-
tions less sensitive to the magnitude of selflessness.
Here we do not investigate further the underlying mecha-
nism guiding this asymmetry, but instead report experiments
conducted in 7 countries that test the robustness of this
asymmetric pattern of reputational inferences across varying
economic and social conditions. Understanding the cross-
cultural robustness of this pattern matters because evalua-
tions of prosociality may be critical for encouraging coop-
eration between unrelated individuals within societies. A
willingness to help others even without the possibility of di-
rect reciprocity is critical for creating the levels of trust and
cooperation necessary for sustaining complex modern soci-
eties and markets (Barclay, 2004; Bowles & Gintis, 2003;
Fehr & Schmidt, 1999; Hamilton, 1964; Trivers, 1971).
Today’s large and relatively impersonal societies make the
close-knit bonds that draw together small communities dif-
ficult to form (e.g., Grossmann & Varnum, 2015; Henrich et
al., 2010; Uskul, Kitayama & Nisbett, 2008).
In lieu of direct reciprocity, the reputational benefits that
come from prosocial actions are thought to create a moti-
vation to behave prosocially, because of a universal desire
to gain social approval (Baumeister & Leary, 1995; Cialdini
& Goldstein, 2004; DeWall & Baumeister, 2006; Twenge
et al., 2007; Willer et al., 2010; Williams, Cheung & Choi,
2000). However, the connection between prosocial behav-
ior and reputational benefits is currently thought to be fairly
straightforward—if a person helps others, then his or her
reputation will benefit commensurably. In contrast, if peo-
ple across cultures fail to differentiate between small and
large prosocial actions, such a result would add important
complexity to existing research.
Understanding how social systems could motivate proso-
cial behavior requires comparing evaluations of prosociality
across cultures. Existing findings on evaluations of proso-
cial actions were obtained from exclusively American sam-
ples, raising concerns that broad conclusions about human
prosociality cannot be drawn due to the idiosyncratic na-
ture of North American cultures (Henrich, Heine & Noren-
zayan, 2010; Nisbett et al., 2001). If prosocial actions are
evaluated differently across cultures, then culture-specific
mechanisms, such as values or norms, may underlie the rep-
utational consequences of prosociality. If, however, proso-
cial actions are evaluated similarly across cultures, then rel-
atively basic and universal mechanisms, such as relative
scope insensitivity, would seem to be guiding reputational
inferences.
Existing research on prosociality across cultures does not
offer a clear prediction about reputational inferences be-
cause it typically focuses on variance in prosocial behavior
rather than in inferences from that behavior. For example,
a recent study examined how people rewarded or punished
others’ prosocial behavior in a repeated public goods game
(Herrmann, Thöni & Gächter, 2008). Whereas participants
in all cultures paid a personal price to punish another per-
son’s selfishness, participants in some cultures also paid a
personal price to punish—instead of reward—another per-
son’s extremely generous behavior. This suggests that some
cultures may not value or admire prosocial behavior, and
may instead disdain generosity. This possibility implies cul-

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
357
tural variability in reputational inferences from prosociality.
In some cultures it may actually hurt one’s reputation to be
really nice. However, punishing extreme prosociality does
not necessarily indicate negative inferences about prosocial
actors. In Hermann et al. (2008), punishing another person’s
generosity could also reflect a strategic attempt to counter
social pressure to contribute to the public pool. More gener-
ally, prosocial behavior can stem from many different mech-
anisms, ranging from admiration of another person to strate-
gic attempts to exploit or manipulate others in specific situ-
ations (Spence, 1973; Zahavi & Zahavi, 1997). Understand-
ing the reputational consequences of prosocial behavior re-
quires measuring the reputational costs and benefits that
emerge in evaluations of others’ prosocial behavior across
the entire spectrum of outcomes, from completely selfish to
completely selfless.
1.1 Overview
Previous research suggested an asymmetry in people’s eval-
uations of another person’s prosocial behavior (Klein & Ep-
ley, 2014). In one experiment (Klein & Epley, 2014, Exp.
3), participants evaluated a person who was given $6 in an
experiment and was offered the opportunity to give some of
it to another participant, with no possibility for reciprocity.
Here we use a similar procedure to test the robustness of
this asymmetry in reputational inferences across 7 different
countries that vary widely in economic and social variables,
and that have also been studied in prior research on proso-
ciality across cultures (Hermann et al., 2008).
Although a person’s reputation may vary along many dif-
ferent dimensions, existing research suggests that a person’s
reputation typically varies only along two fundamental di-
mensions: warmth and competence (Fiske, Cuddy & Glick,
2007; Willis & Todorov, 2006; Wojciszke & Abele, 2008).
Warmth is related to other-oriented outcomes (e.g., friend-
liness, trustworthiness, morality), whereas competence is
more closely related to self-oriented outcomes (e.g., intel-
ligence, talent, skill). Because prosocial actions are more
relevant to one’s treatment of others than for one’s compe-
tence, we predicted that warmth judgments are more likely
to be affected by prosociality, consistent with previous re-
search (Klein & Epley, 2014). Nevertheless, measuring per-
ceived competence enabled us to test whether people view
giving without the possibility of being paid back as a sign of
incompetence.
2 Method
2.1 Participants
We sought to capture cross-cultural variability on several so-
cial and economic dimensions. Our selection of 7 countries
captures variability in social capital, economic prosperity,
democracy and laws, and cultural value dimensions (Table
1). We also selected our cultures to capture variability in the
tendency to reward or high degrees of prosociality as found
in a previous study (Herrmann et al., 2008).
2.2 Procedure
The experiments were conducted between May 2013 and
September 2014. All materials in non-English speak-
ing countries were translated and back-translated to en-
sure semantic accuracy. Austrian participants were re-
cruited through a student email list at the University of
Salzburg. Chinese participants were recruited through the
online panel company Sojump. Danish participants were
recruited through a student email list at Aarhus Univer-
sity. Russian participants were recruited in a classroom
at Novosibirsk State University (n= 73) and through a
psychology students’ email list (n= 122) at Novosibirsk
State University and Novosibirsk State Technical University.
Turkish participants were recruited in a law course at Dogus
University. British participants were recruited through a de-
partmental participant pool at the University of Kent. Amer-
ican participants were recruited via Amazon.com’s M-Turk
online panel. American and Chinese participants, as well as
Russian participants recruited through the students’ email
list were paid nominal amounts. All other participants re-
ceived course credit for participating.
The procedure was identical across all of the experiments.
Participants read about two men who came to a research in-
stitution to participate in a study in which one of them was
given a small amount of money and decided how much of
it to give to another man (as in a “dictator game,” follow-
ing Experiment 3 by Klein & Epley, 2014). We used locally
common names for the actor and referred to the receiver as
“the other person” (see Table 3 for procedural details). Par-
ticipants read the two men had never met each other prior to
the experiment. Participants then read that the giver was free
to decide on any amount to give, from nothing to the entire
endowment. The endowment itself was denominated in the
local currency. To minimize confounds related to the avail-
able endowment, we equated its purchasing power across
cultures to that of 6 American dollars.
We manipulated the amount the giver decided to give to
be either 0, 1/6 of the endowment, 2/6, 3/6, 4/6, 5/6, or the
entire endowment. This manipulation was fully between-
subjects. Behaving equitably by splitting the allocation ben-
efits the self as much as it benefits the other. Giving less than
half benefits the self more than the other person, and so is by
definition relatively selfish. Giving more than half of a finite
endowment benefits the other person more than the self, and
so is by definition relatively selfless. Participants then eval-
uated the giver on traits related to warmth (sincere, warm,
good-natured, caring, tolerant) and competence (competent,
confident, independent, intelligent, competitive; Fiske et al.,

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
358
Table 1: Cultural variability on economic and social dimensions.
Participant pool: Salzburg Online
panel Aarhus Istanbul Novosibirsk Kent Online
panel Sample
average
World
average
Culture: Austria China Denmark Turkey Russia U.K. U.S.A.
Social capital
Share of people who
should be trusted n.a. .55 .67 .16 .24 .29 .36 .38 .28
Norms of civic
cooperation n.a. 9.34 9.27 9.79 8.05 8.65 8.65 8.96 8.64
Economic prosperity
Per capita GDP 50.90 7.30 59.10 11.40 16.80 39.00 53.30 34.00 13.90
Democracy & laws
Rule of law 1.84 −0.49 1.85 0.04 -0.82 1.69 1.60 0.82 0.00
Democracy 16 121 1 81 131 13 14 54 n.a.
Cultural value dimensions
Power distance 80.0 80.0 18.0 66.0 93.0 35.0 40.0 58.9 59.9
Individualism 55.0 20.0 74.0 37.0 39.0 89.0 91.0 57.9 42.8
Masculinity 79.0 66.0 16.0 45.0 36.0 66.0 62.0 52.9 49.8
Uncertainty avoidance 70.0 30.0 23.0 85.0 95.0 35.0 46.0 54.9 67.2
Survival 1.45 −0.61 1.96 −0.35 −1.86 1.37 1.64 0.51 0.06
Traditionalism 0.25 1.16 1.11 −0.83 1.08 0.26 −0.53 0.36 −0.18
In the “Participant pool” column, “Various” refers to online surveys that sampled from various cities and locations
within each country. The social capital variables are from the World Values Survey. The variable “Norms of civic
cooperation” includes three questions from the World Values Survey about disapproval of free-riding by evading paying
taxes or public transit fares; higher values indicate stronger norms. GDP per capita, in unites of $1,000 PPP) is taken
from the International Monetary Fund. The strength of the Rule of Law was taken from the World Bank. Data for
Democracy are ranks of 150 countries taken from the World Audit. Power distance, individualism, masculinity, and
uncertainty avoidance data are taken from Hofstede & Hofstede (2005). Survival (an abbreviation for “Survival(−) vs.
Self-expression(+)” and traditionalism (“Traditionalism(−) vs. Secular-rational(+)”) data are taken from Inglehart &
Norris (2003). The format of this table is similar to Herrmann et al., 2008, Table S1.
2002; Klein & Epley, 2014). All ratings were made on 7-
point scales ranging from 1 (not at all) to 7 (very much).
Our sample sizes appear in Table 3, and are as follows:
Austria, n= 214; China, n= 215; Denmark, n= 181; Russia,
n= 195; Turkey, n= 148; U.K., n= 123; U.S.A., n= 164.
Variation in sample sizes was due to the ease or difficulty
of finding participants in each country. Actual sample sizes
were determined based on time and funding constraints—
we simply collected as many data points as possible under
these constraints.
2.3 Construct validity
To assess construct validity, we conducted principal compo-
nents analyses on the 10 traits we measured in every cul-
ture, using Oblimin rotation (Table 2). In all cultures we
created the warmth composite by averaging the 5 tradition-
ally used traits (sincere, warm, good-natured, caring, toler-
ant) and the competence composite by averaging the other 5
traits (competent, confident, independent, intelligent, com-
petitive).1As Table 2 shows, the warmth and competence
composites generally produced high reliabilities, with one
exception.
1Because “competitive” loaded highly negatively onto the warmth di-
mension, we also conducted the same analyses shown in Tables 4 and 5
after reverse-scoring and incorporating “competitive” into the warmth com-
posite. This does not meaningfully alter any results.

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
359
Table 2: Factor loadings and scale reliabilities of the warmth and competence dimensions across cultures.
Culture
Trait Austria China Denmark Turkey Russia U.K. U.S.A.
Warm .88 −.11 .96 .01 .92 .09 .90 .02 .86 −.03 .92 −.01 .97 .02
Caring .90 −.02 .95 .04 .91 .04 .88 −.01 .87 .08 .91 .05 .96 .04
Good-natured .86 .00 .91 .08 .89 −.10 .88 .05 .89 .04 .94 −.07 .94 .06
Tolerant .78 .19 .96 .00 .85 .11 .75 −.15 .69 .02 .86 .15 .83 .14
Sincere .50 .37 .93 .04 .48 .48 .68 .30 .32 .48 .80 .17 .91 .09
Competent .18 .75 .47 .62 .09 .85 .20 .65 .38 .54 .25 .72 .18 .76
Confident −.22 .82 .37 .69 −.14 .78 −.04 .86 −.22 .79 −.16 .79 −.03 .82
Intelligent .22 .67 .34 .65 .09 .83 .18 .52 .29 .52 .24 .70 .19 .76
Independent .20 .68 .33 .71 −.09 .77 −.10 .75 .00 .79 −.09 .78 −.08 .84
Competitive −.37 .62 −.49 .84 −.74 .40 −.61 .34 −.72 .27 −.75 .38 −.81 .41
αW armth .87 .98 .87 .89 .87 .94 .97
αCompetence .76 .85 .82 .64 .67 .70 .74
The table presents rotated component matrices using Oblimin rotation. The lefthand column in each culture presents
loadings on factor 1 and the righthand column in each culture presents loadings on factor 2. Bolded numbers in the
lefthand column in each culture indicate the traits included in factor 1 (the warmth dimension). Bolded numbers in the
righthand column in each culture indicate the traits included in factor 2 (the competence dimension). The bottom two
rows present the reliabilities (Cronbach’s α) for the result warmth and competence composites for each culture.
3 Results
3.1 Warmth evaluations
We first tested whether prosociality affects reputations
across the entire spectrum of possible actions. For each cul-
ture, we conducted two sets of linear regressions—one with
the amounts of money given predicting warmth evaluations
(Figure 1), and the other with the amount of money given
predicting competence evaluations.
As Table 4 indicates, across all cultures, giving more
leads to a more favorable reputation of warmth. This result
suggests that prosociality has consistent reputational bene-
fits. However, these reputational benefits were largely asym-
metric. We tested this asymmetry using three different anal-
yses of our data, all of which yield similar conclusions.
In the simplest test, we compared evaluations of giving
half of the endowment (3/6) against the most selfish (0/6)
and most selfless (6/6) actions (giving none versus giving
all). As shown in Table 5, participants in all countries judged
the most selfish person more negatively than the fair person;
only in one of the countries (China) was the most selfless
person judged more favorably than a merely equitable per-
son. In 5 out of 7 countries, no statistically reliable differ-
ences emerged in evaluations of the fair person and the most
selfless person, and in one (Turkey) giving the entire amount
was judged more negatively than giving half. We will return
Figure 1: Evaluations of prosociality across cultures.
1 2 3 4 5 6 7
Portion of endowment given
Evaluations of warmth
0 1/6 2/6 3/6 4/6 5/6 6/6
Austria
China
Denmark
Russia
Turkey
U.K.
U.S.A.
to the latter observation in the discussion.
In another test of the asymmetry, we conducted 3 separate
linear regressions for each culture: First, a regression with
all giving amounts predicting warmth; second, a regression
with only selfish actions (giving 3/6 to 2/6 to 1/6 to none of

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
360
Table 3: Procedural details of experiments.
Culture
Austria China Denmark Russia Turkey U.K. U.S.A.
Language German Mandarin Danish Russian Turkish English English
Endowment 6 Euros 60 Renminbi 60 Krones 180 Rubles 12 Liras 6 Pounds 6 Dollars
Name Used for Actor Ben Han Liu Magnus Viktor Murat Bob Mark
Sample Sizes: Giving 0/6 33 31 25 23 22 17 23
Giving 1/6 30 30 26 22 21 16 23
Giving 2/6 28 30 25 30 22 19 25
Giving 3/6 29 31 27 33 21 19 24
Giving 4/6 31 31 26 32 21 18 23
Giving 5/6 30 30 25 34 21 16 23
Giving 6/6 33 32 25 21 20 18 23
Total 214 215 181 195 148 123 164
the endowment) predicting warmth; and third, a regression
with only generous actions (giving 3/6 to 4/6 to 5/6 to all
of the endowment) predicting warmth. Table 4 presents the
results. In all 7 countries, increasingly more selfish actions
led to significantly more negative warmth evaluations. In
contrast, increasingly more selfless actions did not lead to
significantly more positive evaluations of warmth in 5 of the
7 countries. Even in the 2 out of 7 cultures in which increas-
ing generosity led to significantly more positive evaluations,
sensitivity to gradations in selfish actions was higher than
sensitivity to gradations in generous actions. In the U.S., in-
creasing generosity was evaluated more positively (β= .24,
t= 2.33, p= .022), but still not as much as increasing self-
ishness was evaluated more negatively (β= -.71, t= 14.95,
p< .0001), z= 4.33, p< .0001. In China, the sensitivity to
increasingly generous actions was also almost significantly
lower than the sensitivity to increasingly selfish actions, z=
1.76, p= .078.
A final set of analyses sought to better understand the role
of culture in evaluations of prosociality. We therefore tested
whether culture interacts with the magnitude of prosocial
actions. An ANOVA of warmth evaluations on all amounts
given and culture revealed a main effect for amount given,
F(6, 1191) = 187.78, p< .0001, η2
p= .070, a main effect for
culture, F(6, 1191) = 18.06, p< .0001, η2
p= .015, qualified
by an interaction, F(36, 1191) = 3.41, p< .001, η2
p= .093.
To further understand this interaction, we conducted anal-
yses for selfish and generous actions separately. For gener-
ous actions (giving 3/6—6/6 of the endowment), an ANOVA
of warmth evaluations revealed no main effect of amount
given, F(3, 691) = .46, p= .71, a main effect of culture,
F(6, 691) = 25.57, p< .001, η2
p= .090, and an interaction,
F(18, 691) = 2.03, p= .007, η2
p= .050. These results indi-
cate some cultural variation in evaluations of generous ac-
tions. As Table 4 shows, cultural variation in sensitivity to
generous actions emerged from differences in the direction
of the effect of generosity on warmth evaluations. Chinese
and American participants, for example, were most likely
to view greater generosity more favorably (regression βs
= .36 and .24, respectively), whereas Turkish participants
viewed greater generosity more negatively (regression β=
-.22). Participants in other cultures were not sensitive to
magnitude in generous actions, as none of the other relevant
regression βs were statistically significant.
For selfish actions (giving 0–3/6 of the endowment), an
ANOVA of warmth evaluations revealed a main effect of
amount given, F(3, 677) = 175.36, p< .0001, η2
p= .053 a
main effect of culture, F(6, 677) = 6.06, p< .001, η2
p= .025,
and an interaction, F(18, 677) = 2.35, p= .001, η2
p= .059.
These results indicate some cultural variation in evaluations
of selfish actions. As Table 4 shows, in some cultures the
sensitivity to gradations in selfish actions was higher than in
others. American participants, for example, were the most
sensitive to gradations in selfish actions (regression β= .71),
whereas Austrian participants were least sensitive (regres-
sion β= .53). However, unlike generous actions, which had
effects in both directions, increased selfishness always led
to more negative evaluations. Overall, these results indi-
cate that cultural differences were observed both in evalu-
ations of selfish actions and generous actions, but the im-
pact of these cultural differences on evaluations differed be-
tween selfish and generous actions. Whereas greater selfish-
ness always reduced evaluations, greater generosity either
increased, reduced, or did not affect evaluations. We return
to this topic in the Discussion.

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
361
Table 4: Standardized coefficients of regressions reflecting evaluations of prosociality across cultures.
Culture
Trait Austria China Denmark Russia Turkey U.K. U.S.A.
Warmth: All Actions .51* .73* 67* .53* .46* .66* .76*
Warmth: Selfish Actions .53* .54* .65* .60* .65* .65* .71*
Warmth: Generous Actions .00 .36* .05 −.00 −.22* −.00 .24*
Competence: All Actions −.21* .29* −.44* −.11* −.36* −.35* −.06
Competence: Selfish Actions −.00 .00 −.39* .00 −.36* −.41* −.03
Competence: Generous Actions −.00 .19* −.10 −.21* −.00 −.00 −.01
The table presents standardized coefficients of regressions. The “All Actions” row presents regressions of all possible
giving amounts on warmth and competence composites. The “Selfish Actions” row presents regressions of giving
amounts of 0–3/6 of the endowment. The “Generous Actions” row presents regressions of giving amounts of 3/6–6/6
of the endowment. Asterisked coefficients are significant at p< .05.
Overall, these results demonstrate a high degree of simi-
larity in the asymmetry between relatively selfish and gen-
erous actions. Participants’ evaluations of others’ prosocial
actions were consistently more sensitive to gradations in
selfish than in generous actions. These results hold across
cultures that differ markedly on other dimensions between
these seven countries. People, among different cultures, are
generally more sensitive to gradations of selfish behavior
than to gradations of selfless behavior. This consistent pat-
tern was also moderated somewhat by differences across the
cultures we studied. We speculate on the meaning of these
differences amidst the broader similarity we observed in the
Discussion.
3.2 Competence evaluations
The reputational consequences of prosocial behavior were
less clear-cut when examined in terms of competence eval-
uations (Table 4). The composite measure of competence
was consistently less reliable than the composite measure of
warmth across cultures, but we retain the composite because
in most cultures scale reliabilities were acceptable (α> .70;
see Table 2 for details) and to maintain continuity with both
the existing empirical literature and across our samples.
In 5 cultures (Austria, Denmark, Russia, Turkey, U.K.),
greater giving led to significantly lower evaluations of com-
petence across the range of possible outcomes. This result
may have occurred because in our experiments there was
no possibility of reciprocity, which is one of the rationales
for generous giving. Participants may therefore have per-
ceived greater giving as naïve, unwise, or that the person
simply misunderstood the nature of the situation. In the
U.S., greater giving had no statistically reliable relationship
to evaluations of competence. Finally, in China greater giv-
ing led to more favorable competence evaluations. Exam-
ining selfish actions and generous actions separately elimi-
nates most of the statistically reliable relationships between
giving and competence (Table 4). In particular, generous
actions (giving more than half of the endowment) did not
affect competence evaluations, suggesting that participants
did not associate extreme generosity with incompetence.
Overall, giving more does not appear to increase evalua-
tions of competence. If anything, it tends to decrease com-
petence evaluations in this particular context.
4 Discussion
Successful societies require cooperation between unrelated
individuals in order to function effectively. Such prosocial
behavior is encouraged, at least in part, by the reputational
benefits an individual receives from being kind towards oth-
ers and from the reputational costs one incurs when be-
ing unkind towards others. Those who behave prosocially
earn reputations that encourage future trust and cooperation
from others. Those who behave antisocially earn reputa-
tions that create distrust and avoidance. While we cannot
generalize our findings to cultures and subcultures not tested
here, we provide evidence for an asymmetry in these reputa-
tional costs and benefits across 7 cultures. Whereas increas-
ingly selfish actions were judged increasingly negatively
in all cultures we surveyed, increasingly selfless actions—
giving progressively more to others than to the self—were
not judged increasingly positively. In terms of one’s rep-
utation, it pays to be nice, but pays no more to be really
nice. These findings, while not drawing conclusions about
any specific culture, were nevertheless comparable across
cultures that vary on a wide range of social and economic
dimensions. Moreover, prosocial actions also do not earn
reputational benefits in terms of competence evaluations—

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
362
Table 5: Cross-cultural evaluations of giving half of the endowment, giving nothing, and giving the entire endowment.
Culture
Conditions Austria China Denmark Russia Turkey U.K. U.S.A.
Warmth
Giving nothing 3.64 (.96) 3.01 (1.68) 3.23 (1.19) 2.98 (1.08) 2.80 (1.59) 3.58 (.79) 2.97 (1.42)
Giving half 5.16 (1.09) 5.52 (.96) 5.58 (.96) 4.86 (.96) 5.68 (1.26) 5.81 (.87) 6.08 (.69)
Giving all 5.24 (1.08) 6.33 (.75) 5.70 (1.00) 4.76 (1.34) 4.62 (1.77) 5.87 (1.18) 6.43 (.95)
Half vs. nothing t(60) = 5.82* t(60) = 7.20* t(50) = 7.85* t(54) = 6.59* t(41) = 6.57* t(34) = 8.03* t(45) = 9.48*
Half vs. all t(60) = .28 t(61) = 3.39* t(52) = .44 t(52) = -.30 t(39) = -2.21* t(35) = .17 t(44) = 1.43
Competence
Giving nothing 4.68 (.94) 4.51 (1.41) 5.10 (.84) 4.71 (.96) 5.25 (1.30) 5.32 (.66) 5.24 (.90)
Giving half 4.61 (.95) 4.66 (.77) 4.05 (1.14) 4.27 (.80) 3.99 (1.29) 4.47 (.85) 5.18 (.96)
Giving all 4.48 (1.17) 5.46 (1.07) 3.70 (1.26) 4.00 (.96) 3.98 (1.43) 4.63 (.75) 5.21 (.57)
Half vs. nothing t(60) = .79 t(60) = .52 t(50) = -3.73* t(54) = 1.86 t(40) = -3.14* t(34) = -3.28* t(45) = .22
Half vs. all t(60) = .49 t(60) = 3.73* t(52) = 1.06 t(52) = 1.51 t(38) = .03 t(35) = .60 t(44) = .11
The table presents mean warmth and competence evaluations of giving half of the endowment and the entire endowment.
The “difference” rows present independent-samples t-test tests. Asterisks represent significant differences at p< .05.
in Russia, in fact, generosity led to decreased evaluations of
competence.
These results replicate and extend previous findings
among American participants (Klein & Epley, 2014). This
replication therefore addresses concerns about unjustifiably
broad conclusions that could otherwise be drawn from ex-
periments using samples from only a single culture (Hen-
rich, Heine & Norenzayan, 2010). This replication also ad-
dresses recent concerns about the reproducibility of findings
in psychological science (Simmons, Nelson & Simonsohn,
2011), offering 7 additional exact replications of previously
published results.
Despite overarching similarities across cultures in the
asymmetry between evaluations of selfish and generous ac-
tions, potentially interesting cultural differences did emerge
in these results. Participants in some cultures were more
sensitive to gradations of selfish and generous actions than
in others. Understanding why this is the case and how cul-
tural differences in evaluations are related to cultural dif-
ferences in prosocial behavior (e.g., Henrich et al., 2010)
is a productive avenue for future research. For now, we
tentatively raise the possibility that evaluations of proso-
cial actions might be related to anti-social punishment—
decisions to punish extremely prosocial others (Herrmann et
al., 2008). Figure 2 plots anti-social punishment as reported
by Herrmann et al. (2008) along with our participants’ sen-
sitivity to selfish and generous actions (taken from our Ta-
ble 4). Across cultures, anti-social punishment correlates
negatively—but not significantly—with sensitivity to gra-
dations in evaluations of selfish and selfless actions. This
negative correlation could point to an interesting connection
between judgment of others’ prosocial actions and behavior
towards prosocial others. Our ability to test this possibil-
ity is limited because our data contain only 7 cultures, too
small a number to establish meaningful conclusions. Fu-
ture research can measure both evaluations and punishment
decisions to definitively test whether the two are causally
related.
Notwithstanding these possible cultural differences, the
overarching cross-cultural similarities may imply that the
psychological mechanisms underlying asymmetric evalua-
tions of prosocial actions may also be relatively similar
across cultures. These mechanisms may therefore be basic
cognitive or affective processes that are relatively indepen-
dent of culturally conditioned input. Two potential mecha-
nisms have been documented in American samples and are
potential candidates for future investigation. The first is that
people are insensitive to magnitude when evaluating gen-
erous actions because selfish actions are more common and
therefore can be more easily evaluated than generous actions
(Klein & Epley, 2014). Existing research finds that familiar-
ity with a stimulus enables people to notice finer gradations
of this stimulus (Hsee & Zhang, 2010; Morewedge et al.,
2009). The same psychological process can apply in eval-
uations of prosocial actions. The second potential mecha-
nism is the asymmetric affective consequences of prosocial
and selfish actions—generous actions may not increase pos-
itive affect as much as selfish actions of the same magnitude

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
363
Figure 2: Anti-social punishment and evaluations of selfish
(top) and generous (bottom) actions across cultures.
0.0 0.5 1.0 1.5 2.0 2.5
0.50 0.55 0.60 0.65 0.70 0.75
Antisocial punishment from Herrmann et al. (2008)
Sensitivity to gradations in selfish actions
U.S.A.
U.K.
China
Austria
Denmark Turkey
Russia
r(7)=−.20, p=.67 2−tailed
0.0 0.5 1.0 1.5 2.0 2.5
−0.2 0.0 0.2 0.4
Antisocial punishment from Herrmann et al. (2008)
Sensitivity to gradations in generous actions
U.S.A.
U.K.
China
Austria
Denmark
Turkey
Russia
r(7)=−.66, p=.11 2−tailed
increase negative affect (Gray, Ward & Norton, 2014).2Fu-
ture research is needed to test whether these mechanisms
explain the asymmetric pattern of evaluations of prosocial
actions across cultures.
More broadly, the nature of the reputational inferences
we uncover can have important implications for understand-
ing how reputational inferences may motivate prosocial be-
havior. The reputational inferences we have documented
2Another possibility is that people may believe that the motivation of
generous actors are more ambiguous than those of selfish actors, perhaps
because generous actions are seen a non-normative (Miller, 1999). To our
knowledge, no direct evidence for this mechanism currently exists, but it
remains a theoretical possibility.
suggest strong reputational incentives for modestly proso-
cial and cooperative behavior because such behavior pro-
vides the maximum reputational benefit to the actor without
incurring the personal cost of an extremely selfless action.
Regardless of the precise psychological cause of an asym-
metry in evaluations of prosociality, the functional outcome
may be to create social incentives that promote cooperative
behavior. Existing research emphasizes the punishment of
non-cooperators as a necessary mechanism for cooperation
(e.g., Balliet & Van Lange, 2013; Fehr & Schmidt, 1999).
However, the cooperative behavior necessary for sustaining
complex modern societies may also result from the lack of
incentives for very generous prosocial actions, which in turn
incentivizes actors to engage in modestly nice actions. Rep-
utational inferences could nudge societies composed of un-
related individuals into being modestly nice, enabling the
cooperation necessary for successful societies, without hav-
ing to overcome the challenge of motivating people to be
really nice. From an individual’s perspective, behaving
in modestly prosocial ways—but not necessarily extremely
prosocial ways—appears to be the most personally benefi-
cial course of action. It pays for one’s reputation to be nice,
apparently around the globe, but it does not consistently pay
more to be really nice.
References
Balliet, D. & Van Lange, P. A. M. (2013). Trust, pun-
ishment, and cooperation across 18 societies: A meta-
analysis. Perspectives on Psychological Science, 8, 363–
379.
Barclay, P. (2004). Trustworthiness and competitive altru-
ism can also solve the “tragedy of the commons.” Evolu-
tion and Human Behavior, 25, 209–220.
Baumeister, R. F. & Leary, M .R. (1995). The need to be-
long: Desire for interpersonal attachments as a fundamen-
tal human motivation. Psychological Bulletin, 117, 497–
529.
Bowles, S. & Gintis, H. (2003). Origins of human cooper-
ation. In P. Hammerstein (ed.), The genetic and cultural
evolution of cooperation (pp. 429–443). Dahlem Work-
shop Reports: Berlin, Germany.
Buss, D. M. (1989). Sex differences in human mate pref-
erences: Evolutionary hypotheses tested in 37 cultures.
Behavioral and Brain Sciences, 12, 1–49.
Cialdini, R. B., & Goldstein, N. J. (2004). Social influence:
Compliance and conformity. Annual Review of Psychol-
ogy, 55, 591–621.
DeWall, C. N., & Baumeister, R. F. (2006). Alone but feel-
ing no pain: Effects of social exclusion on physical pain
tolerance and pain threshold, affective forecasting, and in-
terpersonal empathy. Journal of Personality and Social
Psychology, 91, 1–15.

Judgment and Decision Making, Vol. 10, No. 4, July 2015 Asymmetric reputations from prosociality across cultures
364
Fehr, E. & Schmidt, K. M. (1999). A theory of fairness,
competition, and cooperation. The Quarterly Journal of
Economics, 114, 817–868.
Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A
model of (often mixed) stereotype content: Competence
and warmth respectively follow from perceived status and
competition. Journal of Personality and Social Psychol-
ogy, 82, 878–902.
Fiske, S. T., Cuddy, A. J. C., & Glick, P. (2007). Universal
dimensions of social cognition: Warmth and competence.
Trends in Cognitive Sciences, 11, 77–83.
Gray, K., Ward, A. F., & Norton, M. I. (2014). Paying it
forward: Generalized reciprocity and the limits of gen-
erosity. Journal of Experimental Psychology: General,
143, 247–254.
Grossmann, I. & Varnum, M. E. W. (2015). Social struc-
ture, infectious diseases, disasters, secularism, and cul-
tural change in America. Psychological Science, 26,
311–324.
Hamilton, W. D. (1964). The genetical evolution of social
behavior. Journal of Theoretical Biology, 7, 1–16.
Henrich, J., Ensminger, J., McElreath, R., Barr, A., Barrett,
C., Bolyanatz, A., Cardenas, J. C., Gurven, M., Gwako,
E., Henrich, N., Lesorogol, C., Marlowe, F., Tracer, D., &
Ziker, J. (2010). Markets, religion, community size, and
the evolution of fairness and punishment. Science, 327,
1480–1484.
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The
weirdest people in the world? Behavioral and Brain Sci-
ences, 33, 61–135.
Herrmann, B., Thöni, C., & Gächter, S. (2008). Antisocial
punishment across societies. Science, 319, 1362–1367.
Hofstede, G. & Hofstede, G. J. (2005). Cultures and Orga-
nizations: Software of the Mind. New York: McGraw-
Hill.
Hsee, C. K., & Zhang, J. (2010). General evaluability the-
ory. Perspectives on Psychological Science, 5, 343–355.
Inglehart, R. & Norris, P (2003). Rising Tide: Gender
Equality and Cultural Change Around the World. Cam-
bridge: Cambridge University Press.
Klein, N. & Epley, N. (2014). The topography of gen-
erosity:zhang Asymmetric evaluations of prosocial ac-
tions. Journal of Experimental Psychology: General,
143, 2366–2379.
Loewenstein, G. F., Thompson, L., & Bazerman, M. H.
(1989). Social utility and decision making in interper-
sonal contexts. Journal of Personality and Social Psy-
chology, 57, 426–441.
Miller, D. T. (1999). The norm of self-interest. American
Psychologist, 54, 1053–1060.
Morewedge, C. K., Kassam, K. S., Hsee, C. K., & Caruso,
E. M. (2009). Duration sensitivity depends on stimulus
familiarity. Journal of Experimental Psychology: Gen-
eral, 138, 177–186.
Nisbett, R. E., Peng, K., Choi, I., & Norenzayan, A. (2001).
Culture and systems of thought: Holistic vs. analytic cog-
nition. Psychological Review, 108, 291–310.
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011).
False-positive psychology: Undisclosed flexibility in data
collection and analysis allows presenting anything as sig-
nificant. Psychological Science, 22, 1359–1366.
Spence, M. (1973). Job market signaling. Quarterly Journal
of Economics, 87, 355–374.
Trivers, R. L. (1971). The evolution of reciprocal altruism.
Quarterly Review of Biology, 46, 35–57.
Twenge, J. M., Baumeister, R. F., DeWall, C. N., Ciarocco,
N. J., & Bartels, J. M. (2007). Social exclusion decreases
prosocial behavior. Journal of Personality and Social
Psychology, 92, 56–66.
Uskul, A. K., Kitayama, S. & Nisbett, R. E. (2008). Eco-
cultural basis of cognition: Farmers and fishermen are
more holistic than herders. Proceedings of the National
Academy of Sciences, 105, 8552–8556.
Veselý, S. (2015). Elicitation of normative and fairness
judgments: Do incentives matter? Judgment and Deci-
sion Making, 10, 191–197.
Willer, R., Feinberg, M., Irwin, K., Schultz, M., & Simp-
son, B. (2010). The trouble with invisible men: How rep-
utational concerns motivate generosity. In S. Hitlin & S.
Vaisey (eds.) The Handbook of the sociology of morality
(pp. 315–330). New York: Springer.
Williams, K. D., Cheung, C. K., & Choi, W. (2000). Cy-
berostracism: Effects of being ignored over the Internet.
Journal of Personality and Social Psychology, 79, 748 –
762.
Willis, J. & Todorov, A. (2006). First impressions: Making
up your mind after 100-ms exposure to a face. Psycho-
logical Science, 17, 592–598.
Wojciszke, B., & Abele, A. E. (2008). The primacy of com-
munion over agency and its reversals in evaluations. Eu-
ropean Journal of Social Psychology, 38, 1139–1147.
Zahavi, A. & Zahavi, A. (1997). The handicap principle: A
missing piece of Darwin’s puzzle. Oxford: Oxford Uni-
versity Press.