Science has long held that people are generally connected to their friends by wealth, popularity or similarity.

But in a recent study investigating the internal mental processes we use to value friendships, Penn psychologists Peter DeScioli and Robert Kurzban found that emotion also plays a critical role in how we prioritize our friends. According to their most recent work, these researchers say that people value friendships based not necessarily on what they believe their friends can do for them, but rather how much they believe their friends care about them.

“Historically, the main theory has been that humans build friendships in order to trade in goods and services,” DeScioli says. “The problem we focused on was that friendship involves more than exchange. People want friends who care about them and do not give just to get something back in return.”

“The issue of what predicts friendships is pretty well known in the networking community,” adds Kurzban. “But what we’ve added here is this notion that not all friendships are created equal.”

In the scientific community, the theory that friends are primarily concerned with each others’ needs rather than the benefits they can get in return for helping is known as the Alliance Hypothesis for Human Friendship. This idea challenges the Theory of Reciprocal Altruism, which states that people are most interested in friendships as a means for exchanges.

In their study, DeScioli and Kurzban examined friend-ranking in the course of three studies involving 285 individuals. Participants were asked to rank their friends, excluding family members and romantic/sexual partners. Later they were asked to rank them again while imagining that their friends would know they were ranked. What the researchers found was that people’s rankings of their 10 closest friends were predicted by their own “perceived rank” among their partners’ other friends.

In all three trials, the results held true even when controlling for factors such as perceived similarity, familiarity and any expected benefits to be gained.

“Friendships are about alliances,” Kurzban says. “The value of an ally, or friend, drops with every additional alliance they must make, so the best alliance is one in which your ally ranks you above everyone else as well.”

If Kurzban and DeScioli are right, and the way we rank our best friends is closely related to how we think they rank us, then the success of popular social networking web sites like Twitter, Facebook and MySpace should not come as a surprise to anyone. People use these platforms daily for gathering information about their friends’ other friendships. The brain stores this information to help us understand the social world not only from our own perspective, but also from the perspectives of our friends, helping us build our alliances.

“We call this ‘friend masking.’ People are being cagy about revealing to people how much more or less they like the different people in their network,” Kurzban says. “If you want people not to leave your alliance, then it’s good for them to think that they have a good chance of attracting you in the case of conflict.”

Applying alliance dynamics to the study of human friendships also might help scientists explain why people are so concerned with social comparisons or with others’ relative superiority in knowledge or skills. Cognitive psychology is also helping explain jealousy and relational aggression among friends–behaviors that are inconsistent with traditional rules governing friendships.

“If you look at the literature on networks, mostly people think of networks as dots connected to other dots,” Kurzban says. “What we’re saying actually is that the strength of those connections matters a lot. I think what we’ll be seeing down the road is more people looking at the strength of these connections and trying to understand these networks in context of alliances.”