Multiplayer games on regular graphs under aspiration-driven imitation: Fixation probabilities and the dilution principle

We investigate how cooperation evolves in multiplayer games on regular graphs of degree k when individuals base their decisions on both their own aspiration levels and the success of their neighbors. We calculate explicit first-order formulas for fixation probabilities using pair approximation and diffusion methods under weak selection. These formulas separate aspiration effects from payoff effects filtered by the graph. We show that increasing the graph degree attenuates the selection effects driven by payoffs. In contrast, the contribution of the aspiration gap Δα=α D-α C is independent of graph degree at first order. We apply the framework to public goods, stag-hunt, and snowdrift games and obtain corresponding conditions for the evolutionary success of cooperation. In the public goods and stag-hunt games, the conditions at large degree retain cost terms, so cooperation requires a sufficiently positive aspiration gap to compensate for the cost of contributing. In the snowdrift game, both benefit and cost contributions vanish in the locally tree-like regime at large degree, leaving the aspiration gap as the leading first-order contribution. Monte Carlo simulations on connected random regular graphs of degree k are consistent with the first-order predictions within their formal regime of validity, while simulations outside this regime are interpreted only as qualitative checks of the predicted trends.