Content uploaded by Ernst Fehr
Author content
All content in this area was uploaded by Ernst Fehr on Dec 15, 2015
Content may be subject to copyright.
Content uploaded by Ernst Fehr
Author content
All content in this area was uploaded by Ernst Fehr on Dec 01, 2015
Content may be subject to copyright.
ARTICLES
Egalitarianism in young children
Ernst Fehr
1,2
, Helen Bernhard
1
& Bettina Rockenbach
3
Human social interaction is strongly shaped by other-regarding preferences, that is, a concern for the welfare of others.
These preferences are important for a unique aspect of human sociality—large scale cooperation with genetic strangers—
but little is known about their developmental roots. Here we show that young children’s other-regarding preferences assume
a particular form, inequality aversion that develops strongly between the ages of 3 and 8. At age 3
–
4, the overwhelming
majority of children behave selfishly, whereas most children at age 7
–
8 prefer resource allocations that remove
advantageous or disadvantageous inequality. Moreover, inequality aversion is strongly shaped by parochialism, a preference
for favouring the members of one’s own social group. These results indicate that human egalitarianism and parochialism
have deep developmental roots, and the simultaneous emergence of altruistic sharing and parochialism during childhood is
intriguing in view of recent evolutionary theories which predict that the same evolutionary process jointly drives both human
altruism and parochialism.
Other-regarding preferences are decisive for the human ability to
achieve and maintain cooperation in large groups of genetic stran-
gers
1,2
. If an individual cares for the welfare of other group members,
he or she is more likely to refrain from free-riding in cooperative
projects. Similarly, if an individual dislikes the free-riding of others—
because it is associated with inequality
3–5
or because it represents a
norm violation
6
—the individual is more likely to punish free-
riders
7–9
. This punishment then constitutes an incentive for potential
free-riders to cooperate. Other-regarding preferences are also
important in public life and politics
10
and they powerfully amplify
reputational incentives in strategic interactions, thus contributing to
the cooperation-enhancing force of reputation opportunities
11–14
.
The developmental origins and proximate mechanisms behind
other-regarding preferences are not well understood, however, despite
recent progress
15–22
. Because we know little about when young chil-
dren start to take the welfare of others into account, we conducted
experiments with 229 young, genetically unrelated, Swiss children
(127 girls and 102 boys) between the ages of 3 and 8. An understanding
of the development of other-regarding preferences in children may
enable us to gain deeper insights into the proximate and ultimate
sources of species differences in preferences and cooperation. The
study of children’s preferences is also of particular interest in light
of recent experiments in non-human primates
23–27
, allowing a more
direct comparison between human and non-human primates.
Experiments with non-human primates have, for example, demon-
strated that chimpanzees show little willingness to provide food to a
familiar conspecific in situations where they could do so with no or
only a small cost
23,24,27
. In view of this result, it is interesting to study
whether (and if so, when) children are willing to provide valuable
resources to their partners. In this way, the large species differences
in cooperation between humans and non-human primates can be
more directly traced back to species differences in other-regarding
preferences.
Investigation of the development of moral judgment
28
and proso-
cial behaviour
29–35
is a rich tradition in psychology, but there is a
surprising lack of studies that isolate the development of other-
regarding preferences from the development of other forms of
prosocial behaviour. The experimental study of other-regarding pre-
ferences in humans involves the conduct of one-shot experiments
with anonymous interaction partners because the behaviour in non-
anonymous face-to-face interactions, or in repeated interactions
with the same partner, can easily be affected by selfish motives. A
subject could, for example, behave prosocially because of the expecta-
tion of future benefits from the partner that accrue as a result of
prosocial behaviour in the current interaction. Selfish motives could
therefore drive prosocial behaviours such as sharing a valuable
resource in non-anonymous face-to-face interactions or in repeated
interactions between the experimental subjects. Measuring other-
regarding preferences without such confounds thus requires the
conduct of anonymous one-shot experiments.
Testing for inequality aversion in children
For this reason, we conducted experiments with young children that
enabled us to measure other-regarding preferences such as inequality
aversion. In the context of our experiments, inequality aversion pre-
vails if subjects prefer allocations that reduce the inequality between
themselves and their partner, regardless of whether the inequality is
to their advantage or disadvantage
3
.
Each subject participated in the three treatments described below
and was paired with one other anonymous partner in each treatment.
Each treatment condition was explained in detail to the decision
maker so that we could be sure that the child had completely under-
stood the experiment and the consequences of the different choices
(see Methods and Supplementary Methods). In all treatments, the
decision maker allocated units of sweets (smarties, jellybabies or
fizzers) to himself and/or to the partner. In the ‘prosocial’ treatment,
which was inspired by recent experiments with chimpanzees
23,24
, the
subject could choose between the allocation (1,1)—that is, 1 for
himself, 1 for partner—and the allocation (1,0). This treatment mea-
sures some elementary form of prosociality, because by choosing
(1,1) the subject can at no cost to himself deliver a benefit to the
partner and, thus, avoid advantageous inequality. In principle, the
choice of (1,1) can be driven by the equality motive
3
or by a motive to
increase the partner’s payoff or both parties joint payoff
36
. Economic
self-interest is not involved in the prosocial game because the
decision maker receives one unit regardless of which choice he makes.
It is therefore also possible that a selfish individual who does not care
about the partner’s payoff will choose (1,1). In fact, because there is
1
University of Zurich, Institute for Empirical Research in Economics, Blumlisalpstrasse 10, CH-8006 Zurich, Switzerland.
2
Collegium Helveticum, Schmelzbergstrasse 25, CH-8092
Zurich, Switzerland.
3
University of Erfurt, Nordha
¨use
r Straße 63, D-99089 Erfurt, Germany.
Vol 454
|
28 August 2008
|
doi:10.1038/nature07155
1079
©2008
Macmillan Publishers Limited. All rights reserved
no reason for a selfish individual to make either choice, a population
of self-interested individuals would choose (1,1) in 50% of the cases.
For this reason, evidence for other-regarding behaviour in the pro-
social game requires that the population of children choose (1,1)
significantly above 50% frequency.
In the ‘envy’ treatment, the subject could choose between (1,1) and
(1,2). Here again, it is possible to deliver a benefit to the partner at no
cost, but the choice (1,2) leads to disadvantageous inequality for the
decision maker. Thus, if an individual just wants to increase the part-
ner’s or the joint payoff, he should choose (1,1) in the prosocial
treatment and (1,2) in the envy treatment. In contrast, if the equality
motive drives behaviour in these two conditions, the subject chooses
(1,1) in both treatments, thus avoiding the unequal allocations (1,0) in
the prosocial treatment and (1,2) in the envy treatment. However, as
in the prosocial treatment, a purely selfish individual has no reason to
make either choice in the envy treatment. For this reason, evidence for
the equality motive in the envy game again requires that the popu-
lation of children choose (1,1) significantly above 50% frequency.
In a third condition, the ‘sharing’ treatment, the subject could
choose between (1,1) and (2,0). This treatment measures a strong
form of inequality aversion because the provision of a benefit for the
partner is costly for the subject. Selfish children should therefore
never make the egalitarian choice in this treatment, indicating that
the choice of (1,1) unambiguously suggests an other-regarding pref-
erence. The sharing treatment also enables us to measure altruism as
defined by evolutionary biology because sharing implies a costly
transfer of a valued resource to another individual.
In addition to these treatments, we also implemented an ingroup
and an outgroup condition ‘across subjects’. In the ingroup con-
dition, the partner came from the same playschool, kindergarten or
school, whereas the partner came from another playschool, kinder-
garten or school in the outgroup condition. The rationale for the
outgroup condition is provided by evidence and theory indicating
that parochialism strongly shapes adult human altruism
37–39
, and that
the same evolutionary process might determine the development of
both human altruism and parochialism
40
.
From self-interest to inequality aversion
Among the 3–4-yr-old children, most behaved selfishly in the
ingroup condition of the sharing game because only 8.7% of the
children were willing to share (Fig. 1). The fact that the frequency
of (1,1) choices did not differ significantly from 50% in the prosocial
game and in the envy game further supports the low incidence of
other-regarding preferences at this age (binomial test; P 5 0.21 for
the prosocial game, P 5 0.68 for the envy game; n 5 23). The preval-
ence of selfish behaviour in the sharing game decreased slightly for
5–6-yr-old children, but 78% were still not willing to share at this age.
Also, as with the 3–4-yr-old children, the frequency of egalitarian
choices in the prosocial and the envy game for the 5–6-yr-olds was
not significantly different from 50% (binomial test; P 5 0.24 for the
prosocial game, P 5 0.41 for the envy game; n 5 36). A substantially
different picture emerged, however, for children at ages 7–8 (Fig. 1):
45% of them showed sharing behaviour, and we also found strong
evidence for other-regarding preferences in the other two games. Of
the 7–8-yr-olds, 78% preferred the egalitarian allocation in the pro-
social game, refuting the null hypothesis of random choices (bino-
mial test, P , 0.001, n 5 56). Similarly, an overwhelming majority
(80%) preferred the egalitarian alternative in the envy game at this
age (binomial test, P , 0.001, n 5 56).
Taken together, the behavioural patterns across all three games sug-
gest that children at age 3–4 show little willingness to share resources
but a non-negligible percentage of the children is willing to make
choices that benefit the recipient if it is not costly. After this age,
other-regarding preferences develop, which take the form of inequality
aversion instead of a simple preference for increasing the partner’s or
the joint payoff. If the motive to increase the partner’s or the joint
payoff were to drive the children’s other-regarding preferences, they
would have then chosen the alternative (1,2) in the envy game. In fact,
however, most of the children at age 7–8 preferred the egalitarian
allocation.
Therefore, if we pool the children’s choices across the various
games, we find that both strongly and weakly egalitarian choices
show a large increase with age (Fig. 2, Supplementary Fig. 3 and
Supplementary Table 1). Egalitarianism, which is characterized by
a (1,1) choice both in the prosocial and the envy game (red columns
in Fig. 2), increases from 21% at age 3–4 to 33% at age 5–6, whereas
60% prefer the egalitarian allocation in both games at age 7–8. The
percentage of egalitarian choices at age 7–8 significantly differs from
an independent random choice in each of both games (binomial test,
P , 0.0001, n 5 56). If the children had made independent, random
choices, only 25% of them would have chosen (1,1) in both the
prosocial and the envy game. If we pool the children’s choices in
all three games, the percentage of children who preferred the egal-
itarian allocation in all three games increases from 4% at age 3–4 to
30% at age 7–8 (Fig. 2). Thus, among those children who choose the
egalitarian allocation in the prosocial and the envy game at age 7–8,
roughly 50% share resources in the sharing game. The other 50%
choose the selfish allocation (2,0) in the sharing game.
It is notable that the share of subjects who maximize the partner’s
payoff by choosing both (1,1) in the prosocial game and (1,2) in the
envy game (blue columns in Fig. 2) decreases sharply from 43% at age
3–4 to 16% at age 7–8. Furthermore, the percentage of subjects who
maximize the partner’s payoff in all three games is only roughly 5%
and does not change much with age. The across-game perspective
also enables us to identify a third type of subject whom we call
‘spiteful’ because they minimize the partner’s payoff in all three
games. The share of spiteful subjects is 22% at age 3–4 and 5–6 and
decreases slightly to 14% at age 7–8, a percentage that is similar to the
relative share of spiteful subjects observed in adult subject pools
41,42
.
Parochial egalitarianism
Parochial tendencies affected children’s choices in all three treat-
ments, and these tendencies are pervasive in the sharing and prosocial
0
20
40
60
80
100
3–4
Age
Prosocial game: (1,1) versus (1,0)
Envy game: (1,1) versus (1,2)
Sharing game: (1,1) versus (2,0)
Percentage of egalitarian choices
5–6 7–8
Figure 1
|
The relative frequency of egalitarian choices across all ingroup
treatments.
In these treatments, the decision maker’s choice determines the
resources of an ingroup partner. The frequency of egalitarian choices
strongly increases with age across all three ingroup treatments, and most
children prefer equality at age 7–8 in the prosocial and the envy game.
However, if equality is costly for the children, they choose the egalitarian
allocation less frequently—as indicated by the behaviour in the sharing
game—and at age 3–4, self-interested choices dominate almost completely.
ARTICLES NATURE
|
Vol 454
|
28 August 2008
1080
©2008
Macmillan Publishers Limited. All rights reserved
games (Fig. 3). The egalitarian choice is roughly 15%–20% more
likely in the prosocial game if the partner is an ingroup member
(Fig. 3a). This difference is highly significant (ingroup dummy in
probit regression, P 5 0.004, z 5 2.92, n 5 229) and a similar
ingroup–outgroup gap prevails across all ages. The largest difference
was found in the sharing game (Fig. 3b), where we observed a strong
increase in the frequency of egalitarian choices if the partner was from
the ingroup (age effect in probit regression, P 5 0.001, z 5 3.33,
n 5 115), whereas the children’s willingness to share even slightly
declined with age in the outgroup condition although this was not
significant (age effect in probit regression, P 5 0.123, z 521.54,
n 5 114). We did observe, however, a highly significant interaction
effect between age and outgroup condition (probit regression,
P , 0.001, z 5 3.50, n 5 229), indicating that the difference between
sharing in the ingroup and outgroup conditions strongly increases
with age. Thus, the children’s altruism and parochialism emerges
simultaneously between the ages of 3 and 8 and is associated with a
very strong ingroup bias at age 7–8 (probit regression, P , 0.001,
z 5 3.58, b 5 105), with very little willingness to share with an out-
group member (only 12% of the children shared in the outgroup
condition at this age).
The prevalence of egalitarian choices in the envy game developed
earlier in the outgroup condition, where the children at age 5–6
already overwhelmingly favour the (1,1) allocation (Fig. 3c). The
willingness to remove disadvantageous inequality towards ingroup
members becomes so prevalent at age 7–8 that an ingroup–outgroup
gap no longer exists. However, averaging across gender hides an
important gender effect in the envy game: boys show much stronger
parochial tendencies than girls do because boys seem to be much less
averse against disadvantageous inequality if the partner is an ingroup
member (Fig. 4a; outgroup dummy in probit regression controlling
for age, P 5 0.001, z 5 3.23, n 5 102). In contrast, girls do not differ-
entiate in their choices between ingroup and outgroup partners
(Fig. 4b; outgroup dummy in probit regression controlling for age,
P 5 0.663, z 520.44, n 5 127), but, like boys, they also show an
increasing trend towards egalitarian choices in the envy game as they
become older.
Birth order and sibling effects
We find a strong ‘only child’ and ‘youngest child’ effect in the sharing
game. Children without siblings showed much more costly sharing
behaviour than children with siblings. On average, children without
siblings were 28% more likely to share than children with siblings—a
highly significant difference (probit regression, P 5 0.006, z 5 2.75,
n 5 197) that also exists if we control for income effects (see
Supplementary Data). With increasing age, however, the difference
between children with and without siblings decreases slightly, as
indicated by an interaction effect between ‘child without siblings’
and ‘age’ in a probit regression (P 5 0.022, z 522.29, n 5 197).
Among the children with siblings, we found that—regardless of
age—the youngest children in a family were 17% less willing to share
than children with younger siblings (probit regression controlling for
age, P 5 0.007, z 5 2.71, n 5 172). Thus, it seems that the mere exist-
ence of siblings or birth order may have an important role in deter-
mining altruistic behaviours. Further analyses and interpretation of
the effects of birth order and other demographic and psychological
characteristics can be found in the Supplementary Information.
0
20
40
60
80
100
Prosocial game:
(1,1) versus (1,0)
a
0
20
40
60
80
100
Sharing game:
(1,1) versus (2,0)
b
0
20
40
60
80
100
Envy game:
(1,1) versus (1,2)
c
Percentage of egalitarian choices
Percentage of egalitarian choices
Percentage of egalitarian choices
3–4
Age
Age Age
5–6 7–8
3–4 5–6 7–8 3–4 5–6 7–8
Figure 3
|
Egalitarian choices across ingroup and outgroup conditions.
Grey dashed lines denote when the partner was an ingroup member; black
solid lines represent when the partner was an outgroup member (
a–c). a,In
the prosocial game, the children remove inequality that favours themselves
more often if the partner is an ingroup member.
b, Egalitarian choices
slightly decrease over time in the sharing game (with very little sharing at age
7–8) if the partner is an outgroup member, whereas sharing with ingroup
members strongly increases with age, providing strong evidence for
parochial altruism in children.
c, In the envy game, children develop a
preference for equality much earlier if the partner is an outgroup member
but eventually the aversion against disadvantageous inequality with regard
to ingroup members becomes so strong that ingroup–outgroup differences
are small.
4%
14%
30%
17%
19%
30%
4%
6%
5%
39%
22%
11%
22%
22%
14%
0
20
40
60
80
100
Strongly egalitarian Weakly egalitarian
Strongly generous Weakly generous
Spiteful
3–4
Age
5–6 7–8
Percentage of relative frequency
Figure 2
|
Behavioural types in the ingroup condition. The figure classifies
subjects according to their behaviour in all three games, that is, in the
prosocial game ((1,1) versus (1,0)), the sharing game ((1,1) versus (2,0)), and
the envy game ((1,1) versus (1,2)). Strongly egalitarian subjects choose the
egalitarian allocation in all three games. Weakly egalitarian subjects choose
the egalitarian allocation only in the prosocial and the envy games, but not in
the sharing game, where egalitarian behaviour is costly. Strongly generous
subjects choose the allocation that maximizes the partner’s payoff in all three
games. Weakly generous subjects maximize the partner’s payoff only in the
prosocial and the envy games, but not in the sharing game, where generous
behaviour is costly. Spiteful subjects choose the allocation that minimizes
the partner’s payoff in all three games. The percentage of egalitarian subjects
increases steeply with age, whereas the share of generous subjects declines.
Moreover, most subjects who are willing to share at age 7–8 belong to the
egalitarian and not to the generous type of subjects.
NATURE
|
Vol 454
|
28 August 2008 ARTICLES
1081
©2008
Macmillan Publishers Limited. All rights reserved
Discussion
The development of inequality aversion relatively early in childhood
is particularly interesting in the light of ethnographic evidence that
suggests that there is a strong role of egalitarian ‘instincts’ in human
evolutionary history. There is considerable ethnographic evidence
that egalitarian concerns have shaped many human small-scale soci-
eties
43,44
. For example, sharing of large hunted game across families
seems to have been the rule rather than the exception in small-scale
societies, and egalitarian sentiments also play a part in contemporary
large-scale societies
4,5
. The important role of egalitarian sentiments in
human evolutionary history raises the possibility that there may have
been cultural or even genetic transmission that favours egalitarian
behaviours. In fact, recent evidence from behavioural genetics sug-
gests that egalitarian behaviour in the ultimatum game has a genetic
component
45
.
The simultaneous development of altruistic behaviour and paro-
chialism and the gender differences in parochialism are also interest-
ing in view of evolutionary theories that predict that the same
evolutionary process
40
jointly determines human altruism and paro-
chialism, meaning that these traits co-evolve in such a way that either
both or neither of them evolves. According to the theory, the driving
force behind this evolutionary process is frequent intergroup
conflict. Because mainly males were involved in intergroup fights, it
seems possible that evolution favoured a gender bias in parochialism.
In fact, a payoff advantage relative to the outgroup may have been
particularly advantageous for males because it strengthened the
ingroup’s position in intergroup conflicts. Males bore the main cost
of intergroup conflict in terms of injuries and deaths and often gained
more than females in the case of victory because of the increase in the
pool of potential mating partners
46
. Thus, evolution may have
favoured a greater sensitivity in males for payoff advantages relative
to outgroup members. In view of this prediction, we find it notable
that boys showed a much stronger ingroup bias than girls in the envy
game.
These potential evolutionary roots of human egalitarianism and
parochialism do not preclude culture and socialization from playing
an important part in other-regarding preferences; they may even be a
main factor in their evolution
2,47
. As the children move from an
informal playgroup to kindergarten and then on to formal schools,
they may learn that equality is a rule that the authorities (for example,
the teachers) endorse. Thus, the children probably acquire some of
the normative rules of the society which surrounds them during the
age period on which we focused
30
. In this context, it is interesting that
the motive to increase the partner’s payoff declined strongly between
ages 3 and 8, whereas egalitarian behaviour strongly increased in this
age period (Fig. 2), providing a hint about the content of the norm-
ative rule they acquired. The children may also become more sens-
itive with regard to the opinions of others about themselves, a
cognition which requires the ability to understand that one’s actions
affect what other people believe about oneself. Recent evidence indi-
cates that adult humans care even about what anonymous others may
think about them
48,49
. Theory-of-mind and perspective-taking abil-
ities are certainly conducive to such social cognitions. Therefore, if
older children care about what anonymous others think about them,
they may be more prone to behave in a normatively appropriate way
because—owing to their age—they are more likely to have theory-of-
mind and perspective-taking abilities
30,50
.
Our results also indicate that important inter-species differences
occur in other-regarding preferences when compared to the patterns
observed in chimpanzees
23,24,27
and marmosets
26
. In the prosocial
game, adult chimpanzees, who could allocate food in a face-to-face
interaction to a familiar recipient, did not show a significant pref-
erence for the (1,1) choice
23,24,27
, whereas most of the children
develop a preference for the egalitarian choice in this game even
though their partner is anonymous. Furthermore, as the sharing
game indicates, many children at age 7–8 are also willing to share
with an anonymous ingroup member. The facts in the envy game also
contrast sharply with experimental findings in both chimpanzees and
marmosets. The children developed a widespread aversion against
disadvantageous inequality, whereas chimpanzees have been shown
to be unwilling to take actions to remove inequality between them-
selves and a conspecific
24
, and marmosets even tend to take actions
that generate disadvantageous inequality for themselves if the action
provides food for the partner
26
. In view of the decisive role of other-
regarding preferences for a species ability to achieve large-scale coop-
eration, the observed inter-species differences are probably import-
ant in explaining why humans are so exceptionally cooperative.
METHODS SUMMARY
Each child played all three games against anonymous partners. To avoid satiation
effects, three different currencies (smarties, jellybabies and fizzers) were used in
randomized order in the three games. In each game, subjects had two mutually
exclusive choices that were represented with two cardboards; we drew two circles
with arrows on each (see Supplementary Fig. 1). One arrow pointed to the
decision maker, illustrating that the sweets in that circle goes to him or her,
whereas the other arrow pointed to a group photo which had been made earlier.
In the ingroup condition, the group photo showed the members of the child’s
playgroup, kindergarten or school, whereas it showed the members of a different
playgroup, kindergarten or school in the outgroup condition. The photos were
used to communicate the partner’s ingroup or outgroup status to the children in
the game. Depending on whether the ingroup or the outgroup condition applied,
the decision maker was told that the sweet(s) in the other circle (if there were any
0
20
40
60
80
100
a
0
20
40
60
80
100
b
Percentage of egalitarian choicesPercentage of egalitarian choices
A
g
e
3–4 5–6 7–8
Figure 4
|
Gender differences in parochial egalitarianism in the envy game
where the child could choose between (1,1) and (1,2). a
, b, The propensity
of boys (
a) and girls (b) for egalitarian choices in the ingroup (grey dashed
line) and outgroup (black solid line) condition. The difference between
ingroup and outgroup condition is large for boys but virtually absent
for girls.
ARTICLES NATURE
|
Vol 454
|
28 August 2008
1082
©2008
Macmillan Publishers Limited. All rights reserved
in that circle) would be given to one of the children on the ingroup photo or
outgroup photo, respectively. If there was no sweet in the circle, it was made clear
to the child that the choice of the corresponding cardboard indicates that the
partner would receive nothing. We also controlled for the spatial assignment of
the different alternative. In the prosocial game, for example (shown in
Supplementary Fig. 1), the alternative (1,1) was randomly assigned to the left
or right side. Thus, preferences for left or right cannot explain any of our find-
ings. We also made it clear to the children that neither other children nor their
parents or teachers will be informed about their decisions (see Supplementary
Methods).
Full Methods and any associated references are available in the online version of
the paper at www.nature.com/nature.
Received 12 March; accepted 6 June 2008.
1. Bowles, S. Microeconomics: Behavior, Institutions and Evolution 93
–
126 (Princeton
Univ. Press, 2004).
2. Boyd, R. T. & Richerson, P. The Origin and Evolution of Cultures (Oxford Univ. Press,
2005).
3. Fehr, E. & Schmidt, K. M. A theory of fairness, competition, and cooperation. Q. J.
Econ. 114, 817
–
868 (1999).
4. Fowler, J. H., Johnson, T. & Smirnov, O. Human behaviour: Egalitarian motive and
altruistic punishment. Nature 433, E1 (2007).
5. Dawes, C. T. et al. Egalitarian motives in humans. Nature 446, 794
–
796 (2007).
6. Falk, A. & Fischbacher, U. A theory of reciprocity. Games Econ. Behav. 54, 293
–
315
(2006).
7. Fehr, E. & Ga
¨
chter, S. Altruistic punishment in humans. Nature 415, 137
–
140
(2002).
8. Gu¨rerk, O., Irlenbusch, B. & Rockenbach, B. The competitive advantage of
sanctioning institutions. Science 312, 108
–
111 (2006).
9. Hauert, C. et al. Via freedom to coercion: The emergence of costly punishment.
Science 316, 1905
–
1907 (2007).
10. Fong, C., Bowles, S. & Gintis, H. in Handbook on the Economics of Giving, Reciprocity,
and Altruism (eds Kolm, S.-C. & Ythier, J. M.) (Elsevier, 2006).
11. Milinski, M., Semmann, D. & Krambeck, H. J. Reputation helps solve the ‘tragedy
of the commons’. Nature 415, 424
–
426 (2002).
12. Brown, M., Falk, A. & Fehr, E. Relational contracts and the nature of market
interactions. Econometrica 72, 747
–
780 (2004).
13. Nowak, M. A. & Sigmund, K. Evolution of indirect reciprocity. Nature 437,
1291
–
1298 (2005).
14. Rockenbach, B. & Milinski, M. The efficient interaction of indirect reciprocity and
costly punishment. Nature 444, 718
–
723 (2006).
15. Harbaugh, W. T. & Krause, K. Children’s altruism in public good and dictator
experiments. Econ. Inq. 38, 95
–
109 (2000).
16. de Quervain, D. J.-F. et al. The neural basis of altruistic punishment. Science 305,
1254
–
1258 (2004).
17. Sutter, M. Outcomes versus intentions: On the nature of fair behavior and its
development with age. J. Econ. Psychol. 28, 69
–
78 (2007).
18. Sutter, M. & Kocher, M. G. Trust and trustworthiness across different age groups.
Games Econ. Behav. 59, 364
–
382 (2007).
19. Benenson, J. F., Pascoe, J. & Radmore, N. Children’s altruistic behavior in the
dictator game. Evol. Hum. Behav. 28, 168
–
175 (2007).
20. Fliessbach, K. et al. Social comparison affects reward-related brain activity in the
human ventral striatum. Science 318, 1305
–
1308 (2007).
21. Cesarini, D. et al. Heritability of cooperative behavior in the trust game. Proc. Natl
Acad. Sci. USA 105, 3721
–
3726 (2008).
22. Ellingsen, T. & Johannesson, M. Anticipated verbal feedback induces altruistic
behavior. Evol. Hum. Behav. 29, 100
–
105 (2008).
23. Silk, J. B. et al. Chimpanzees are indifferent to the welfare of unrelated group
members. Nature 437, 1357
–
1359 (2005).
24. Jensen, K., Hare, B., Call, J. & Tomasello, M. What’s in it for me? Self-regard
precludes altruism and spite in chimpanzees. Proc. Biol. Sci. 273, 1013
–
1021 (2006).
25. Jensen, K., Call, J. & Tomasello, M. Chimpanzees are rational maximizers in an
ultimatum game. Science 318, 107
–
109 (2007).
26. Burkart, J. M., Fehr, E., Efferson, C. & van Schaik, C. P. Other-regarding preferences
in a non-human primate: Common marmosets provision food altruistically. Proc.
Natl Acad. Sci. USA 104, 19762
–
19766 (2007).
27. Vonk, J. et al. Chimpanzees do not take advantage of very low cost opportunities
to deliver food to unrelated group members. Anim. Behav. 75, 1757
–
1770 (2008).
28. Turiel, E. in Handbook of Child Psychology Vol. 3 (eds Damon, W. & Eisenberg, N.)
863
–
932 (Wiley and Sons, 1998).
29. Knight, G. P. & Kagan, S. Development of prosocial and competitive behaviors in
anglo-american and mexican-american children. Child Dev. 48, 1385
–
1394
(1977).
30. Eisenberg, N. & Mussen, P. H. The Roots of Prosocial Behavior in Children
(Cambridge Univ. Press, 1989).
31. Thompson, C., Barresi, J. & Moore, C. The development of future-oriented
prudence and altruism in preschoolers. Cogn. Dev. 12, 199
–
212 (1997).
32. Eisenberg, N. & Fabes, R. A. in Handbook of Child Psychology Vol. 3 (eds Damon, W.
& Eisenberg, N.) 701
–
778 (Wiley and Sons, 1998).
33. Murnigham, J. K. Ultimatum bargaining by children and adults. J. Econ. Psychol. 19,
415
–
445 (1998).
34. Warneken, F. & Tomasello, M. Altruistic helping in human infants and young
chimpanzees. Science 311, 1301
–
1303 (2006).
35. Warneken, F. et al. Spontaneous altruism by chimpanzees and young children.
PLoS Biol. 5, e184 (2007).
36. Charness, G. & Rabin, M. Understanding social preferences with simple tests.
Q. J. Econ. 117, 817
–
869 (2002).
37. Goette, L., Huffman, D. & Meier, S. The impact of group membership on
cooperation and norm enforcement: Evidence using random assignment to real
social groups. Am. Econ. Rev. 96, 212
–
216 (2006).
38. Bernhard, H., Fischbacher, U. & Fehr, E. Parochial altruism in humans. Nature 442,
912
–
915 (2006).
39. Shinada, M., Yamagishi, T. & Ohmura, Y. False friends are worse than bitter
enemies: ‘‘Altruistic’’ punishment of in-group members. Evol. Hum. Behav. 25,
379
–
393 (2004).
40. Choi, J. K. & Bowles, S. The coevolution of parochial altruism and war. Science 318,
636
–
640 (2007).
41. Van Lange, P. A. M. The pursuit of joint outcomes and equality in outcomes: An
integrative model of social value orientation. J. Pers. Soc. Psychol. 77, 337
–
349
(1999).
42. Falk, A., Fehr, E. & Fischbacher, U. Driving forces behind informal sanctions.
Econometrica 73, 2017
–
2030 (2005).
43. Boehm, C. Egalitarian behavior and reverse dominance hierachy. Curr. Anthropol.
34, 227
–
254 (1993).
44. Boehm, C. Impact of the human egalitarian syndrome on darwinian selection
mechanics. Am. Nat. 150, S100
–
S121 (1997).
45. Wallace, B., Cesarini, D., Lichtenstein, P. & Johannesson, M. Heritability of
ultimatum game responder behavior. Proc. Natl Acad. Sci. USA 104, 15631
–
15634
(2007).
46. Chagnon, N. A. Life histories, blood revenge, and warfare in a tribal population.
Science 239, 985
–
992 (1988).
47. Henrich, J. et al. ‘‘Economic man’’ in cross-cultural perspective: behavioral
experiments in 15 small-scale societies. Behav. Brain Sci. 28, 795
–
855 (2005).
48. Dana, J., Cain, D. M. & Dawes, R. M. What you don’t know won’t hurt me: Costly
(but quiet) exit in dictator games. Organ. Behav. Hum. Decis. Process. 100, 193
–
201
(2006).
49. Ellingsen, T. & Johannesson, M. Paying respect. J. Econ. Perspect. 21, 135
–
149
(2007).
50. Wellman, H. M., Cross, D. & Watson, J. Meta-analysis of theory-of-mind
development: The truth about false belief. Child Dev. 72, 655
–
684 (2001).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements This paper is part of the Research Priority Program
‘Foundations of Human Social Behaviour—Altruism versus Egoism’ at the
University of Zurich and of the Swiss National Competence Center in research on
affective sciences, which is financed by the Swiss National Science Foundation. We
also thank N. Kessler for her research assistance during the conduct of the
experiments.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. Correspondence and requests for materials should be
addressed to E.F. (efehr@iew.uzh.ch) or B.R. (bettina.rockenbach@uni-erfurt.de).
NATURE
|
Vol 454
|
28 August 2008 ARTICLES
1083
©2008
Macmillan Publishers Limited. All rights reserved
METHODS
Subjects. We recruited young children (3–8-yr-old) from playschools, kinder-
gartens and first grades for our experiments. We received permission for the
experiments from the school president, the school board, the teachers and the
parents. Of the parents that we asked, 92% agreed so that a total of 229 children
participated in the study: 62 children from playschools, 75 from kindergartens
and 92 children from grade one. The children live in Rapperswil-Jona, a small
city in the canton of St Gallen, Switzerland. Two women aged between 20 and 25
conducted the experiment. The experimental instructions are reproduced in the
Supplementary Information.
Experimental procedures. Each child played the prosocial game, the envy game
and the sharing game against anonymous partners. The order of the games was
counterbalanced across subjects. The partners came either from the ingroup
(same playschool, kindergarten or school) or from an outgroup (a different
playschool, kindergarten or school).
Payoffs. We needed an experimental currency desirable for the younger and
older children in our sample. Therefore we decided to use various sweets. To
avoid satiation effects, a different currency was used in each of the three games
(smarties, jellybabies or fizzers), and these were randomized across games. Before
the experiment started, we asked each child whether he or she likes those sweets
(all of them did). At the end of all three treatments we asked each child whether
they liked all three sweets the same or whether they liked one more than the
other. If they indicated that they liked some of the sweets more than the others we
asked which they liked the most and least (see Supplementary Information). The
average ‘liking rates’ of the different currencies were identical. Moreover, we
asked the parents to rate on a 7-point scale how much their children value the
sweets and their answers showed that our currencies provide equally strong
incentives for the children across all ages.
The choice situation. To ensure that the children could easily understand the
choice problem, we used a set up that made the two available choices transparent.
The two mutually exclusive choices were represented with two cardboards; we
drew two circles with arrows on each (see Supplementary Fig. 1). One arrow
pointed to the decision maker, illustrating that the sweet in that circle goes to him
or her, whereas the other arrow pointed to a group photo which had been made
earlier. In the ingroup condition, the group photo showed the members of the
child’s playgroup, kindergarten or school, whereas it showed the members of a
different playgroup, kindergarten or school in the outgroup condition. The
photos were used to communicate the partner’s ingroup or outgroup status to
the children in the game. Depending on whether the ingroup or the outgroup
condition applied, the decision maker was told that the sweet(s) in the other
circle (if there were any in that circle) would be given to one of the children on the
ingroup photo or outgroup photo, respectively. If there was no sweet in the
circle, it was made clear to the child that the choice of the corresponding card-
board implies that the partner would receive nothing. The advantage of this
design is that the same procedure can be used for the ingroup and the outgroup
condition, and that the children can grasp the partner’s ingroup or outgroup
status very easily. In fact, many children across all ages immediately expressed
their knowledge of the children on the photo, that is, they spontaneously indi-
cated that they knew or did not know the children in the photo. On the basis of
this spontaneous insight, it was then easy to explain that the partner will be a
member of the child’s group or the member of another group.
We also controlled for the spatial assignment of the different alternatives. In
the prosocial game, for example, which is shown in Supplementary Fig. 1
(Supplementary Methods), the alternative (1,1) was randomly assigned to be
on the left or the right side. Thus, preferences for left or right cannot explain any
of our findings. We also made it clear to the children that neither other children
nor their parents/teachers would be informed about their decisions.
Before a child played a game we ensured that he or she fully understood the
game situation (that is, the available choices, the implications of different choices
for the allocation of currencies for ‘self’ and ‘partner’, the partner’s ingroup–
outgroup status, and so on). The subjects had to answer several questions for this
purpose (Supplementary Methods). Only three children had problems in
answering these questions, which shows that we successfully implemented the
three games even for the youngest participants in the sample. The three children
who could not answer correctly were excluded from the data analyses (229 is the
number of children who correctly answered the questions). Once the children
had answered the questions correctly, they were asked to make a decision in the
first game before the second game was presented and explained. Because we
ensured that the children understood the payoff implications of the two available
choices very well and because it was clear that the choices were mutually exclusive
(that is, only one cardboard could be chosen), the children did not make mis-
takes while choosing, for example, by indicating that they wanted to choose both
cardboards. Nor did any children reverse their opinion during the choice pro-
cess, that is, children who first chose one cardboard but later switched to the
other.
Questionnaire. After all children had participated in the experiment we sent the
parents a questionnaire in which we asked them about characteristics of their
child, such as whether he or she can easily imagine how other children feel
(‘empathy’), whether there are siblings, the birth order, who primarily cares
for the child during the day, the number of regular playmates, and so on. We
sent 198 questionnaires to the parents and 161 questionnaires were sent back.
The questionnaire is reprinted in the Supplementary Information.
doi:10.1038/nature07155
©2008
Macmillan Publishers Limited. All rights reserved
