Developing the sharing sense

10 minute read

Following on after yesterday’s post about hunter-gatherer population structure, I ended with the proposal that cooperation may be a “cognitive technology” in the same way suggested for numbers (“Number as cognitive technology”).

The technology perspective attracts me. It seems a productive way to examine the interaction between innate and extrinsic factors leading to human behaviors. We learn about numbers. Without a development of the brain within a cultural setting with widespread counting and training in number use, people don’t develop the habits of mind that allow rapid comparison of cardinal values. They can still operate on sets of objects and compare their quantities, but they are missing a shorthand, a symbolic shortcut, that comes with learning and practice. Numerical concepts, invented and repeatedly used by human societies, give learners access to this symbolic method of problem-solving.

Cooperation and other prosocial behaviors are similar in some respects. Whether you share with another person or not in a particular concept depends on the rules about sharing that you learned as a member of your society. What’s interesting is that these rules change with age in various ways. So I went looking in the developmental psychology literature for some data about how kids share. My notes here are just a start – and I’m pretty sure they’re rough to read near the end – but I found it interesting how the data seem to illuminate the issue of cooperation in the archaeological record.


Toddlers can, in some circumstances, exhibit a surprising degree of understanding about the intentions of others. They can also be surprisingly helpful – that is, they can see when another individual wants something, and can actively help that other person to get it. A paper last fall by Kristen Dunfield and colleagues Dunfield:2010 gives a nice review of this kind of helping behavior in toddlers aged 18 and 24 months.

Replicating previous work by Warneken and Tomasello (2006, 2007), we found that by 18 months, infants are beginning to identify the situations in which helping behavior is required; that is, they will aid instrumentally by retrieving an item that is out of a persons reach, thus fulfilling anothers unmet goal. Further, the present study found a similar frequency of helping behavior to Warneken and Tomasello (2006), even though in the current study participants only received one experimental helping trial as opposed to the three trials they received in the previous paradigm. In light of previous studies, helping behavior may also be seen as young as 14 months, though the contexts in which it occurs are less flexible, owing perhaps to an emerging understanding of goal-directed activities (Warneken & Tomasello, 2007), recognition of the means by which certain unmet goals can be fulfilled, and the physical ability to mediate the completion of the goal.

However, as I well remember from my own toddlers, the “prosocial” characteristics of infants can be temperamental, to say the least. Dunsfield and colleagues considered 18 and 24-month-olds, finding substantial heterogeneity among individuals in the kind of helping or sharing behavior they exhibited.

While acknowledging the dangers of arguing from a null effect, it is the case that although the majority of the participants engaged in at least some prosocial behavior, there were no correlations between the various prosocial behaviors. Further, the most common pattern of response was to engage in only one type of prosocial behavior (helping or sharing). Although the tendency to engage in prosocial behavior in general tended to increase across our two timepoints, the increase was not the result of systematic development within or between the various subtypes of prosocial behavior. Thus, we have no evidence in the present study for across the board prosocial behavior within individuals in these two age groups. With future research that explores the consistency both within and between the multiple specific types of behavior, and that considers enduring behavior over time in a longitudinal manner (Eisenberg et al., 1999), it may be the case that helping, comforting, and sharing do not cluster together within an individuals repertoire and perhaps should not be grouped together as one general category of unified behavior in infancy.

A natural question is, what does it take to manage any kind of sharing at all among children this young? By this age most children have experienced thousands of times when an adult or another caregiver has performed the opposite role, giving the child what she cannot reach herself. This long history of positive exemplars for sharing and cooperative behavior nevertheless leaves substantial variation among children in how they actually behave in a similar context.

The first article by Warneken and Tomasello cited above Warneken:Tomasello:2006 compared human children with chimpanzee juveniles of a similar age. They showed that the human children did show these prosocial tendencies by 18 months, but that so do chimpanzees – at least to a certain extent. The chimpanzee juveniles handled the most indexical of the tasks relatively well – the case where a person is reaching for something but needs help to reach it. Other tasks didn’t bring out the cooperative nature in chimpanzee juveniles:

However, the chimpanzees did not help the human reliably in the other types of tasksthat is, in those involving physical obstacles, wrong results, or wrong means. In a follow-up study, we gave them two additional tasks of these typesdesigned to make the human's problem especially salient and with more time for a responseand they still did not help in these tasks (14). Presumably, when someone is reaching with an outstretched arm toward an object, the goal is in principle easier to understand and the kind of intervention follows straightforwardly. This could explain why out-of-reach tasks (in contrast to the other scenarios) elicited more helping by children and the only instances of helping by chimpanzees. Children and chimpanzees are both willing to help, but they appear to differ in their ability to interpret the other's need for help in different situations.

This goes some distance toward explaining what children need to make them potential helpers. They need some way of figuring out the goal of the person who needs help, and they need to have no goal of their own that directly conflicts. Before Warneken and Tomasello’s work, chimpanzee juveniles had not shown signs of such prosocial behaviors in other experimental contexts. Those authors attribute the difference to food: Most chimpanzee experiments had involved food treats, attempting to get individuals to share food with each other. The chimpanzee’s own desire for the food may directly interfere with the goals of other individuals – a conflict that is hardly likely to lead to sharing, even in human toddlers.

There is little sense in calling the chimpanzee behavioral pattern “rudimentary”, as psychologists sometimes do. The human pattern here is rudimentary compared to the extent of helping and sharing that occur later in childhood. The human children in this context seem to have an ability to diagnose the intentions of another individual more than do the chimpanzees. They also seem to have more patience for helping, in some sense. Warneken and Tomasello returned to the topic in a 2009 review Warneken:Tomasello:2009 that puts forward the situation with respect to sharing, helping, and information transfer. They note that human language depends on cooperation in a way that chimpanzee vocalizations do not. It may not be coincidental that language is learned across the same ages as cooperative behaviors.

Preschool-aged children

Olson and Spelke Olson:Spelke:2008 reported on a slightly more intricate study with 3.5-year-old children. They assessed sharing behavior in which children had to divide a pool of items among a number of recipients. These potential recipients sometimes included both relatives and strangers. In other instances, the potential recipients varied in terms of whether they had interacted with the children by sharing with them. Olson and Spelke intended to find whether children of this age would engage in direct and indirect reciprocity, and whether they would skew their distribution of the resource toward relatives as opposed to strangers.

What they found is that kids of this age typically divy things up fairly:

Children may have distributed resources equally on the four-resource trials for either of two reasons. First, it is possible that children will resort to equal sharing whenever resources are plentiful and will favor family, friends, reciprocators, and generous others only under conditions of scarcity. Such a possibility is consistent with the finding that social conflicts among older children and adults arise primarily when resources are limited ([Jackson, 1993] and [Sherif et al., 1961]). Alternatively, the equality response may be driven by a predisposition to distribute resources in a one-to-one correspondence with recipients whenever such a distribution is possible. That predisposition, in turn, could arise either spontaneously or through the internalization of an explicit rule children are taught by parents and other adults.

As soon as they can manage matching objects with people, they are parceling out things one to a person. That’s obviously an integral part of most children’s experience – everything from passing out parts in a game, to passing out food at dinner. So the behavior itself is highly reinforced if not explicitly taught, and it may well be explicitly taught to most children.

The children in Olson and Spelke’s trials also tended to share more with people who had previously been generous in the past, either directly or indirectly to the child. By rewarding past generosity, the children were fulfilling their end of a reciprocity arrangement. This seems pretty relevant to the dynamics in ancient human groups; if a 3-year-old can manage the basics of reciprocity, it may not have taken much to push people into a stable hunting and gathering economy, which is based on reciprocity.

School-age children

Here’s what interested me the most. Kids at 3.5 years already get the idea of sharing equally and fairly. So you might think this would be deeply ingrained in older children. But instead what we see is that older children start to reason more and more like adults, which ironically makes them share less evenly. They just get more clever about how to rationalize their choice to be unfair.

For example, a nice study by Gummerum and colleagues Gummerum:2008 compared students age 9 to age 17 for their performance in the “dictator game.”

The “dictator game” is an experimental model that has been repeatedly employed in adults to study the themes of cooperation and altruism. An individual is given control over how to divide a single sum between herself and another anonymous person. The individual can choose any division down all for himself and zero for the anonymous player.

Gummerum and colleagues added a twist, making individuals work in groups of three to decide on their offers. The offers then reflected not only the preferences of individuals going into the study but also their moral reasoning with each other after discussing the offers in small groups. This yielded an interesting, almost ethnographic picture of how the children came to make their decisions about appropriate offers.

They found that the offers made by groups were strongly influenced by the level of moral reasoning employed by group members. When a student who favored a low offer was arguing at a higher level of sophistication, the group was more likely to adopt a low offer. And vice-versa – when the clever student was arguing for a more equitable offer at a higher level, the group was more likely to give more. Girls gave higher offers than boys in the experiment as well.

In a game like this, the sharing and reciprocity aspects of prosocial behavior are transformed into moral questions. No punishment befalls students who choose to make low offers in the dictator game; yet there is the consideration of self-regard. And others have heard the arguments that a student makes, affecting her reputation. Moral reasoning is, in other words, public.

Concluding thoughts

What I find so interesting about comparing children of different ages, is not about cooperation but instead about how the rules are shifted to higher levels of description. Sharing and reciprocity are quite simple, and children can manage them young, although irregularly. Kids can learn about sharing and helping in a rather unsophisticated way, and their performance reflects very simple expectations. Equal division, turn-taking, and punishment of defectors are all integral parts of early childhood.

Obviously, any humans living in foraging societies in the recent past have grappled similarly with the moral aspects of cooperation and altruism. But that moral reasoning comes at an age far past when children are taught about the importance of fairness, sharing and helping. The kind of dynamic that concerns many anthropologists – how do foraging peoples maintain the rules that underlie reciprocity and altruistic behavior – is simply at a different level than the dynamic that actually inculcates cooperation. Yet with children who learn systematically to help and cooperate, such behaviors have a much higher chance of existing stably, even in small societies. If there is any cognitive invention that a human society would not want to lose, I think some conception of fairness may be it.