Original article
Children's altruistic behavior in the dictator game

https://doi.org/10.1016/j.evolhumbehav.2006.10.003Get rights and content

Abstract

This study examined developmental and socioeconomic status (SES) differences in young children's altruistic behavior in the dictator game (DG). Children aged 4, 6, and 9 years old from six British primary schools played the DG with genetically unrelated individuals using stickers as resource. Results demonstrated that older children and children from higher SES environments behaved more altruistically, although the majority of children displayed altruistic behavior even at the youngest age level. Results buttress conclusions based on studies from diverse cultures and from brain imaging research by providing additional evidence for the fundamental nature of altruistic behavior, as well as for the probable influence of local socialization practices on development.

Introduction

Virtually all developmental studies conclude that young children behave selfishly towards genetically unrelated individuals (for reviews, see Damon, 1977, Eisenberg & Fabes, 1998). A few studies report the development of altruistic behavior during middle childhood (ages 6–12 years), but evidence remains inconclusive due to discrepancies obtained in relation to variations in methods and populations (Eisenberg & Fabes, 1998). Simple economic games, such as the dictator game (DG), provide an avenue for systematically examining altruistic behavior even in young children.

Compared to all other species, human beings display much higher levels of altruistic behavior towards genetically unrelated individuals (for a review, see Haviland, Prins, Walrath, & McBride, 2004). Individuals display altruistic behavior towards nonkin across diverse societies ranging from small subsistence communities containing familiar individuals to large market-based societies with anonymous exchanges (Henrich et al., 2005). Whereas altruism between kin can be explained in terms of genetic benefits (Hamilton, 1964), the mechanisms underlying cooperation or altruism between genetically unrelated individuals remain obscure. Examining the extent to which humans attain their own individual goals while simultaneously displaying altruistic behavior towards unrelated individuals constitutes a powerful lens with which to illuminate knowledge of human social interactions.

Over the past decade, there has been remarkable progress in developing methods for isolating variables that influence human social interaction. Contrary to prior emphasis on the overwhelming importance of self-interest in interactions with nonkin, newer research highlights the critical importance of a number of unselfish behaviors (for reviews, see Camerer, 2003, Roth, 1995). Research from diverse cultures demonstrates that social norms for conditional altruism exist (e.g., Axelrod, 1984). Individuals who obey social norms for altruistic behavior obtain rewards, whereas those who defy norms incur punishment even at a cost to those inflicting the punishment (for a review, see Fehr & Fischbacher, 2004). Although large absolute differences in norms for altruistic behavior occur across cultures, all cultures appear to reward altruistic behavior and to punish selfishness to some extent (Henrich et al., 2005, Henrich et al., 2006). Furthermore, recent neural (Fehr & Rockenbach, 2004) and hormonal (Kosfeld, Heinrichs, Zak, Fischbacher, & Fehr, 2005) studies suggest an evolved basis for the maintenance of altruistic norms.

Much of the research documenting the ubiquitous nature of norms for altruistic behavior rests on the development of simple economic games that permit the isolation of specific social behaviors (for reviews, see Camerer, 2003, Roth, 1995). Although these games elicit criticism (as lacking in ecological validity, hence constituting imperfect predictors of actual social behavior), the strength of these games lies in their ability to isolate and test simple predictions without naturally occurring confounds that obfuscate the influence of any one variable. Furthermore, field evidence of naturally occurring altruistic behavior both provides ecological validity for and permits modification of the parameters of the games so that they more closely mimic elements of natural interactions (Falk & Fehr, 2003).

Games vary in the ability of other players to reciprocate, the number of players, the number of iterations, the ability of outside observers to reward and punish selfish behavior, payoffs from cooperation versus defection, and the reputation of other players (for a review, see Fehr & Fischbacher, 2003). They also differ in terms of stable individual differences of players, such as their age, sex, and culture (Henrich et al., 2005). The creation of positive-versus-negative feedback conditions simulating specific market conditions permits researchers to examine increasingly realistic and complex economic decisions (Camerer & Fehr, 2006).

While recent research on the display of altruistic behavior investigates its biological basis (e.g., Fehr & Rockenbach, 2004, Kosfeld et al., 2005), as well as its pervasiveness and variability across diverse cultures (Henrich et al., 2005), the development of altruistic behavior in children remains virtually unexplored. A number of researchers consequently have concluded that research into the development of altruistic behavior in children would aid in illuminating the nature of altruism (Fehr & Fischbacher, 2003, Fehr & Fischbacher, 2004, Henrich et al., 2005).

Only a few developmental studies have utilized paradigms in which children must distribute resources. In a study by Harbaugh, Krause, and Liday (2003), 7-, 9-, 10-, 14-, and 18-year-olds played the DG, as well as other economic games. In the DG, each participant was given 10 tokens (each of which, they were told, was worth about $0.25) and asked to divide them between the participant and another anonymous classmate. At the end of the study, older participants exchanged the tokens for money, and younger ones exchanged the tokens for toys or school supplies. Envelopes were used to ensure the confidentiality of proposers' choices and to avoid responses designed to please adults. Results showed that dictator proposals increased with age. Based on the figures provided, average dictator proposals were approximately 0.5 tokens for 7-year-olds, 1.70 tokens for 9-year-olds, and almost 4 tokens for 18-year-olds. Girls also behaved slightly more altruistically than boys. Researchers utilizing other methods have similarly reported small developmental increases in altruistic behavior in Western societies in contexts in which children must specifically distribute resources, but the results are mixed (for reviews, see Eisenberg & Fabes, 1998, Murnigham & Saxon, 1998).

These same studies also report a slight tendency for girls to behave more prosocially than boys in public situations (Eisenberg & Fabes, 1998). With the exception of the Harbaugh et al. (2003) study, however, researchers have not utilized the DG. Results from economic games with adults yield inconsistent sex differences, with females behaving more altruistically in some studies (e.g., Eckel & Grossman, 1998, Eckel & Grossman, 2001) but not in others (e.g., Gurven, 2004, Henrich et al., 2005). Field studies of adults in Western societies however, show that men, more than women, demonstrate prosocial behavior towards genetically unrelated individuals (for a review, see Eagly & Crowley, 1986).

Finally, Henrich et al. (2005) conclude from their study of economic games across 15 small-scale societies that the degree of market integration exerts the strongest influence on altruistic behavior towards nonkin. In societies in which all transactions occur with familiar individuals over short expanses of time, the requirement for altruistic behavior towards anonymous individuals remains minimal. The opposite holds true in societies that are reliant on large market-based economies. Thus, Henrich et al. infer that adults in societies that vary in their degree of market integration likely socialize their children to display differing levels of altruistic behavior towards anonymous others. Studies comparing children in impoverished versus more advantaged subcultures in Western societies demonstrate that children from impoverished subcultures are less likely to trust others or to expect others to reciprocate altruistic behavior (for an extensive review, see Evans, 2004). Likewise other studies have demonstrated a link between the way adults earn a living and their child-rearing values in industrial (Kohn, 1990) and small-scale societies (Barry, Child, & Bacon, 1959).

The strengths of the DG rest on its uniform procedure, simplicity, and wide application, which permit systematic comparisons of altruistic behavior across diverse individuals and contexts. In the DG, one individual (the proposer) dictates how much the other individual (the recipient) will gain (for detailed descriptions, see Kagel & Roth, 1995). Unlike other games, the recipient does not have the power to reject the offer. The key components of the DG consist of a one-shot allocation of a valued resource from one proposer to one recipient, both of whom are anonymous to each other and to others. For convenience, resources generally are divided into discrete parts, such as 10 units, which then can be divided between the two players. DGs vary in the degree to which the researcher is aware of the proposer's allocation, the recipient, and whether a rationale is provided for sharing (e.g., Hill & Gurven, 2004). Most DGs have been played with university students who typically allocate, on average, 20–30% of their resources to the recipient (e.g., Forsythe, Horowitz, Savin, & Sefton, 1994). Of all the economic games developed to date, only the DG is simple enough to represent a valid measure of altruistic behavior in young children.

Based on prior findings using economic games with adults and based on the research by Harbaugh et al. (2003), this study was designed to examine developmental, socioeconomic, and gender effects on children's altruistic behavior in the DG. Pilot testing demonstrated that the youngest age at which children could understand the rules of the game was 4 years. Efforts to use more complex economic games, such as ultimatum or public goods games, with this age group failed due to the inability of young children to comprehend other players' perspectives (for a review, see Wellman, Cross, & Watson, 2001).

Four-year-old, 6-year-old, and 9-year-old children were included to examine as early as possible altruistic behavior and developmental changes during early and middle childhood years. Because formal schooling commences in England at the age of 4 years, schools provide an obvious venue for examining altruistic behavior towards genetically unrelated yet familiar others. The specific variables of interest were: the age during which children would exhibit altruistic behavior, whether this would depend on the context in which they were raised [lower or higher socioeconomic status (SES)], and whether sex differences in altruistic behavior occur in young children. Children in schools where most parents live in government-provided council housing because they cannot afford their own homes were defined as coming from lower SES environments. Children who attend schools where most parents own their own homes were defined as coming from higher SES environments. Parents who cannot afford their own homes likely participate less in wage labor and other anonymous delayed exchanges and, therefore, likely socialize their children differently than parents of children who participate in the labor market (Barry et al., 1959, Henrich et al., 2005, Kohn, 1990).

For the resource, a large variety of highly attractive stickers were used. Stickers were selected because they are highly valued by children of these ages and are used by teachers in all participating schools to reward outstanding behavior. All stickers were purchased outside the country to enhance their value by ensuring that no child had previously obtained an identical one. Although it is possible that children from lower SES environments with fewer resources valued the stickers more (e.g., Bruner & Goodman, 1947), this is unlikely because children in all six schools received stickers for superior performance. Thus, stickers did not represent a scarcer resource for children from lower SES environments versus children from higher SES environments.

Section snippets

Participants

Children from six classes each at the 4-, 6-, and 9-year-old grade levels in four low SES primary schools situated next to public housing and children from four classes each at the 4-, 6-, and 9-year-old grade levels in two higher SES primary schools served as proposers. The SES of the school is defined in England by the number of children receiving free lunches. In lower SES schools, >50% of children in each school received free lunches. In higher SES schools, <5% of children in each school

Results

An analysis of variance (ANOVA) performed on the number of stickers donated to classmates, with age level, SES, sex of proposer, and sex of recipient as independent variables, yielded significant effects of Age Level [F(2, 336)=3.02, p=.05], SES [F(1, 336)=7.33, p<.007], and Age Level×Sex of Proposer×Sex of Recipient [F(2, 336)=3.95, p<.02]. None of the other effects was significant [SES×Sex of Recipient, F(1, 336)=2.74, ns; Sex of Recipient, F(1, 336)=2.57, ns; Age Level×Sex of Recipient, F(1,

Discussion

Developmental findings from the present study indicate that, even by 4 years of age, a majority of children from both lower and higher socioeconomic environments behave altruistically by donating at least one sticker to an anonymous classmate. This indicates an early basis for altruistic behavior. With increasing age, children behave more altruistically, particularly those from more privileged environments. By 9 years, compared to children from lower SES environments, children from higher SES

Acknowledgments

Thanks to all schools and children who participated, and to Frank Marlowe for comments on an earlier draft on this manuscript. The first author expresses her gratitude also to Richard Wrangham and the Department of Biological Anthropology at Harvard University.

References (42)

  • J.S. Bruner et al.

    Value and need as organizing factors in perception

    Journal of Abnormal and Social Psychology

    (1947)
  • C. Camerer

    Behavior game theory: Experiments in strategic interaction

    (2003)
  • C. Camerer et al.

    When does “economic man” dominate social behaviour?

    Science

    (2006)
  • A. Campbell

    Staying alive: Evolution, culture, and women's intra-sexual aggression

    Behavioral and Brain Sciences

    (1999)
  • W. Damon

    The social world of the child

    (1977)
  • A.H. Eagly et al.

    Gender and helping behaviour: A meta-analytic review of the social psychological literature

    Psychological Bulletin

    (1986)
  • C.C. Eckel et al.

    Are women less selfish than men? Evidence from dictator experiments

    Economic Journal

    (1998)
  • C.C. Eckel et al.

    Chivalry and solidarity in ultimatum games

    Economic Inquiry

    (2001)
  • N. Eisenberg et al.

    Prosocial development

  • G.W. Evans

    The environment of childhood poverty

    American Psychologist

    (2004)
  • E. Fehr et al.

    The nature of human altruism

    Nature

    (2003)
  • Cited by (389)

    • The Good of Rules: A pilot study on prosocial behavior

      2023, Journal of Behavioral and Experimental Economics
    View all citing articles on Scopus
    View full text