Traditionally, game theorists posit that cooperation rests on theories of reciprocity, and therefore our confidence that if we cooperate with others, that they will do likewise with us.  In the early days of file-sharing platforms, the platforms would often state how many other users were currently sharing files in a bid to bolster this social proof.

The problem was, often these stats were not entirely accurate.  Research from the University of Pennsylvania explores whether such misleading tactics can actually support cooperation or whether they undermine the trust required to work well with others.

Building trust

The researchers developed a mathematical model to simulate the creation and sustaining of trust within a community.  The model suggests that a degree of deceit can actually promote cooperative activity within the community.

At the heart of the model was the notion that humans are, at heart, cooperative, but that the threshold at which someone will cooperate differs for each individual.  Some are quite happy to cooperate even if few others are doing so, whereas others would rather wait until there are more signs of cooperative behavior from the crowd.

Depending on the cooperation thresholds for the various members of the community, the group will likely have either a high or low level of cooperation when looked at collectively.  The researchers’ aim was to understand how obfuscation may manipulate things upwards or downwards.

During the simulations, the researchers assumed that each individual would join the community as a blank slate, with an inherent belief that the other members would cooperate.  This led to them typically cooperating too.

Tipping point

There came a point, however, where those individuals would gain experience and begin to understand just how much cooperation was really going on, which, depending on their own rates of cooperation, would see them continue to cooperate, freeload, or leave the group.

When the real cooperation rate was kept secret for longer, and therefore the learning rate was slowed down, this resulted in cooperation levels becoming higher, with those savvier individuals seeming to leave the group.

“And because those savvy individuals are the ones that don’t cooperate as readily, that leaves only the individuals who are cooperating, so the average rate of cooperation gets very high,” the researchers say.

Dominant behavior

Similarly, the researchers suggest that cooperative behavior could grow to become the norm provided there was a constant flow of naive individuals into the community.  The researchers believe that their findings mark a clear deviation from our traditional understanding of cooperation.

“Typically when we and others have considered how to maintain cooperation, it’s been thought that it’s important to punish cheaters and to make that public to encourage others to cooperate,” they say. “But our study suggests that a side effect of public punishment is that it reveals how much or how little people are cooperating, so conditional cooperators may stop cooperating. You might be better off hiding the cheaters.”

The researchers plan to follow up their original study with further experiments with human volunteers to test whether the mathematical modeling works in the real world as well as it does in the mathematical world.

“You can see how conditional cooperation factors into behavior during this pandemic, for example,” they conclude. “If you think a lot of people are being careful (for example, wearing masks and social distancing), you might as well, but if the expectation is that not many people are being careful you may choose not to. Mask wearing is easy to observe, but other behaviors are harder, and that affects how the dynamics of these behaviors might unfold.”