The Selective Trust

You’d expect that knowing your opponent would change how you play. Friends should cooperate more. Strangers should defect more. Social proximity should be a general modifier on strategic behavior.

Maroo and Vemuri test this across three games — Dictator, Ultimatum, and Centipede — with participants in no-tie, weak-tie, and strong-tie relationships, eliciting both behavior and beliefs. The result: social proximity has no significant effect on offers in the Dictator and Ultimatum games. Strong-tie and no-tie players behave identically when the interaction follows norms.

But in the Centipede Game — a sequential game where success requires predicting your partner’s specific future moves — strong-tie dyads cooperate longer, defect later, and align their beliefs to their partners more accurately. Social knowledge enters the decision only when the game structure makes it necessary.

The through-claim: relationships don’t globally bias strategic behavior. They provide information that the decision-maker deploys selectively, based on whether the game requires partner-specific reasoning. In norm-driven games (split this pie fairly), everyone follows the same script regardless of relationship. In games requiring mental modeling of a specific other (will they cooperate on move 3?), relational knowledge becomes cognitively relevant.

This reconciles conflicting prior research — some studies find social proximity effects, others don’t. The disagreement was never about social proximity. It was about the games.

Trust is not a bias. It’s a tool, used when the structure demands it.