Preference Evolution under Partner Choice

Abstract

We study preference evolution when agents choose whom to interact with. In the short run, subjective preferences influence both partner choice and strategic behavior, which affect their material payoffs. These payoffs, in turn, determine how preferences evolve in the long run. To model this "match-to-interact" process, we combine stable matching with correlated equilibrium, allowing matched agents to coordinate on self-enforcing agreements. We show that evolution under endogenous matching need not favor either purely selfish or purely efficient preferences. Instead, we identify two classes of preferences that combine these concerns nonlinearly and can prevail because they sustain efficient play while protecting agents from exploitation. Under incomplete information, however, only one of these classes retains its evolutionary advantage. This class combines efficiency with parochialism, a form of group-level selfishness that strengthens incentives for self-sorting.