Original title: Cognitive constraints govern collective outcomes in strategic network formation
Hicham DAHMOU, Frédéric Moisan, Jean-Claude Dreher · September 2026
People build social networks through personal choices, but they cannot see the whole network to know if a connection is truly worth it. The researchers created a computer model to see how this limited information affects the entire group. In the model, agents pay to form links based on a restricted local view, random choices, and optimism about what lies beyond their sight. When decision noiseRandomness in choices, so agents do not always pick the best option even when they can see it. is low and optimism is in a moderate sweet spot, agents build near-efficient networks from an empty start. They need enough optimism to reach out, but not so much that they fail to cut unnecessary ties. This sweet spot shrinks in larger groups, though wider visibility helps. Moving first or receiving early links leads to becoming a central hub. Finally, an agent who builds without caring about costs helps large groups but hurts small ones.
You are organizing a block party and can only see the houses on your own street. You know other streets exist, but not which families would get along, so you invite partly on hope. Too cautious, the party fizzles; too generous, you keep adding people and never drop the ones who do not get along. In this paper, agents face the same choice with connections: using only local information and a bit of optimism, they form or cut links, and the whole network's success depends on that balance.
Real social and professional networks form through people who see only their own ties, yet everyone depends on the wider structure. This model shows that such limited, optimistic decisions can still produce networks with high overall value—but only within a narrow band of behavior, and the band shrinks as groups get larger. That matters for designing platforms, teams, and markets that nudge local choices toward collective benefit.
The abstract reports a computational model, not experiments with human participants, so it does not show that real people actually form networks this way. It also does not establish how much optimism or visibility real groups have, or where the favorable regime sits in practice. The single payoff-ignoring agent's benefit appears only in larger groups, so the result does not generalize to all group sizes.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Cognitive constraints govern collective outcomes in strategic network formation — Hicham DAHMOU, Frédéric Moisan, Jean-Claude Dreher — PsyArXiv — doi:10.31234/osf.io/9nkcb_v2
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.